gcc/ChangeLog:
[official-gcc.git] / gcc / ChangeLog
blobac9a89f7ba882bad0f2ab5a7ff7a26e0539a6877
1 2017-01-26  Martin Sebor  <msebor@redhat.com>
3         * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
4         HAVE_DFmode before using XFmode or DFmode.
5         (parse_directive): Avoid using the z length modifier to avoid
6         the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
8         PR middle-end/78703
9         * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
10         to accept adjustment as an array.
11         (get_int_range): New function.
12         (struct directive): Make width and prec arrays.
13         (directive::set_width, directive::set_precision): Call get_int_range.
14         (format_integer, format_floating): Handle width and precision ranges.
15         (format_string, parse_directive): Same.
17 2017-01-26  Jakub Jelinek  <jakub@redhat.com>
19         PR debug/79129
20         * dwarf2out.c (generate_skeleton_bottom_up): For children with
21         comdat_type_p set, just clone them, but keep the children in the
22         original DIE.
24         PR debug/78835
25         * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26         which have direct callers with -fvar-tracking-assignments enabled
27         in the current TU.
28         (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
29         inside of type units.
31 2017-01-26  Martin Sebor  <msebor@redhat.com>
33         PR middle-end/78703
34         * gimple-ssa-sprintf.c (struct result_range): Add likely and
35         unlikely counters.
36         (struct format_result): Replace number_chars, number_chars_min,
37         and number_chars_max with a single member of struct result_range.
38         Remove bounded.
39         (format_result::operator+=): Adjust.
40         (struct fmtresult): Remove bounded.  Handle likely and unlikely
41         counters.
42         (fmtresult::adjust_for_width_or_precision): New function.
43         (fmtresult:type_max_digits): New function.
44         (bytes_remaining): Handle likely and unlikely counters.
45         (min_bytes_remaining): Remove.
46         (format_percent): Simplify.
47         (format_integer, format_floating): Set likely and unlikely counters.
48         (get_string_length, format_character, format_string): Same.
49         (format_plain, should_warn_p): New function.
50         (maybe_warn): Call should_warn_p.  Update diagnostic messages
51         and handle those for all directives, including plain strings.
52         (format_directive): Handle likely and unlikely counters.
53         Remove unnecessary quoting from diagnostics.  Add an informational
54         note.
55         (add_bytes): Remove.
56         (pass_sprintf_length::compute_format_length): Simplify.
57         (try_substitute_return_value): Handle likely and unlikely counters.
59 2017-01-26  Carl Love  <cel@us.ibm.com>
61         * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
62         bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
64 2017-01-26  Vladimir Makarov  <vmakarov@redhat.com>
66         PR target/79131
67         * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
68         endianess for subregs into account.
69         * lra-constraints.c (lra_constraints): Do risky transformations
70         always on the first iteration.
71         * lra-lives.c (check_pseudos_live_through_calls): Add arg
72         last_call_used_reg_set.
73         (process_bb_lives): Define and use last_call_used_reg_set.
74         * lra.c (lra): Always continue after lra_constraints on the first
75         iteration.
77 2017-01-26  Kirill Yukhin  <kirill.yukhin@gmail.com>
79         * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
80         constant.
81         * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
83 2017-01-26  Jakub Jelinek  <jakub@redhat.com>
85         * config/i386/avx512fintrin.h (_ktest_mask16_u8,
86         _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
87         * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
88         _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
89         * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
90         __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
91         OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
92         * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
93         (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
95 2017-01-26  Marek Polacek  <polacek@redhat.com>
97         PR c/79199
98         * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
99         for the third operand.
101 2017-01-26  Jakub Jelinek  <jakub@redhat.com>
103         PR middle-end/79236
104         * omp-low.c (struct omp_context): Add simt_stmt field.
105         (scan_omp_for): Return omp_context *.
106         (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
107         context to the _simt_ SIMD stmt.
108         (lower_omp_for): For combined SIMD with sibling _simt_
109         SIMD, make sure to use the same decls in _looptemp_
110         clauses as in the sibling.
112 2017-01-26  David Sherwood  <david.sherwood@arm.com>
114         PR middle-end/79212
115         * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
116         all contexts.
118 2017-01-26  Jakub Jelinek  <jakub@redhat.com>
120         PR target/70465
121         * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
122         emit fld b; fld a; if possible.
124         * brig-builtins.def: Update copyright years.
125         * config/arm/arm_acle_builtins.def: Update copyright years.
127 2017-01-25  Michael Meissner  <meissner@linux.vnet.ibm.com>
129         PR target/79179
130         * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
131         constraint instead of o for the stxsd instruction.
133 2017-01-25  Carl Love  <cel@us.ibm.com>
135         * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
136         of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
138 2017-01-25  Jonathan Wakely  <jwakely@redhat.com>
140         * doc/invoke.texi (C++ Dialect Options): Fix typo.
142 2017-01-25  Richard Biener  <rguenther@suse.de>
144         PR tree-optimization/69264
145         * target.def (vector_alignment_reachable): Improve documentation.
146         * doc/tm.texi: Regenerate.
147         * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
148         and add a comment.
149         * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
150         earlier changes with respect to TYPE_USER_ALIGN.
151         (vector_alignment_reachable_p): Likewise.  Improve dumping.
153 2016-01-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
155         PR target/79145
156         * config/arm/arm.md (xordi3): Force constant operand into a register
157         for TARGET_IWMMXT.
159 2016-01-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
161         * doc/invoke.texi (-fstore-merging): Correct default optimization
162         levels at which it is enabled.
163         (-O): Move -fstore-merging from list to...
164         (-O2): ... Here.
166 2017-01-25  Richard Biener  <rguenther@suse.de>
168         PR debug/78363
169         * omp-expand.c: Include debug.h.
170         (expand_omp_taskreg): Make sure to generate early debug before
171         outlining anything from a function.
172         (expand_omp_target): Likewise.
173         (grid_expand_target_grid_body): Likewise.
175 2017-01-25  Maxim Ostapenko  <m.ostapenko@samsung.com>
177         PR lto/79061
178         * asan.c (get_translation_unit_decl): New function.
179         (asan_add_global): Extract modules file name from globals
180         TRANSLATION_UNIT_DECL name.
182 2017-01-24  Eric Botcazou  <ebotcazou@adacore.com>
184         PR target/77439
185         * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
186         for long calls with APCS frame and VFP.
188 2017-01-24  David Malcolm  <dmalcolm@redhat.com>
190         * cfg.c (original_copy_tables_initialized_p): New function.
191         * cfg.h (original_copy_tables_initialized_p): New decl.
192         * cfgrtl.c (relink_block_chain): Guard the call to
193         free_original_copy_tables with a call to
194         original_copy_tables_initialized_p.
195         * cgraph.h (symtab_node::native_rtl_p): New decl.
196         * cgraphunit.c (symtab_node::native_rtl_p): New function.
197         (symtab_node::needed_p): Don't assert for early assembly output
198         for __RTL functions.
199         (cgraph_node::finalize_function): Set "force_output" for __RTL
200         functions.
201         (cgraph_node::analyze): Bail out early for __RTL functions.
202         (analyze_functions): Update assertion to support __RTL functions.
203         (cgraph_node::expand): Bail out early for __RTL functions.
204         * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
205         __RTL functions.
206         * function.h (struct function): Update comment for field
207         "pass_startwith".
208         * gimple-expr.c: Include "tree-pass.h".
209         (gimple_has_body_p): Return false for __RTL functions.
210         * Makefile.in (OBJS): Add run-rtl-passes.o.
211         * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
212         accessor.
213         (gcc::pass_manager::get_clean_slate): New accessor.
214         * passes.c: Include "insn-addr.h".
215         (should_skip_pass_p): Add logging.  Update logic for running
216         "expand" to be compatible with both __GIMPLE and __RTL.  Guard
217         property-provider override so it is only done for gimple passes.
218         Don't skip dfinit.
219         (skip_pass): New function.
220         (execute_one_pass): Call skip_pass when skipping passes.
221         * read-md.c (md_reader::read_char): Support filtering
222         the input to a subset of line numbers.
223         (md_reader::md_reader): Initialize fields
224         m_first_line and m_last_line.
225         (md_reader::read_file_fragment): New function.
226         * read-md.h (md_reader::read_file_fragment): New decl.
227         (md_reader::m_first_line): New field.
228         (md_reader::m_last_line): New field.
229         * read-rtl-function.c (function_reader::create_function): Only
230         create cfun if it doesn't already exist.  Set PROP_rtl on cfun's
231         curr_properties.  Set DECL_INITIAL to a dummy block.
232         (read_rtl_function_body_from_file_range): New function.
233         * read-rtl-function.h (read_rtl_function_body_from_file_range):
234         New decl.
235         * run-rtl-passes.c: New file.
236         * run-rtl-passes.h: New file.
238 2017-01-24 Jeff Law  <law@redhat.com>
240         * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
241         buffer size.
243 2017-01-24  Bin Cheng  <bin.cheng@arm.com>
245         PR tree-optimization/79159
246         * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
247         (record_nonwrapping_iv): Improve boundary using above function if no
248         value range information.
250 2017-01-24  Pekka Jääskeläinen  <pekka@parmance.com>
251             Martin Jambor  <mjambor@suse.cz>
253         * brig-builtins.def: New file.
254         * builtins.def (DEF_HSAIL_BUILTIN): New macro.
255         (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
256         (DEF_HSAIL_SAT_BUILTIN): Likewise.
257         (DEF_HSAIL_INTR_BUILTIN): Likewise.
258         (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
259         * builtin-types.def (BT_INT8): New.
260         (BT_INT16): Likewise.
261         (BT_UINT8): Likewise.
262         (BT_UINT16): Likewise.
263         (BT_FN_ULONG): Likewise.
264         (BT_FN_UINT_INT): Likewise.
265         (BT_FN_UINT_ULONG): Likewise.
266         (BT_FN_UINT_LONG): Likewise.
267         (BT_FN_UINT_PTR): Likewise.
268         (BT_FN_ULONG_PTR): Likewise.
269         (BT_FN_INT8_FLOAT): Likewise.
270         (BT_FN_INT16_FLOAT): Likewise.
271         (BT_FN_UINT32_FLOAT): Likewise.
272         (BT_FN_UINT16_FLOAT): Likewise.
273         (BT_FN_UINT8_FLOAT): Likewise.
274         (BT_FN_UINT64_FLOAT): Likewise.
275         (BT_FN_UINT16_UINT32): Likewise.
276         (BT_FN_UINT32_UINT16): Likewise.
277         (BT_FN_UINT16_UINT16_UINT16): Likewise.
278         (BT_FN_INT_PTR_INT): Likewise.
279         (BT_FN_UINT_PTR_UINT): Likewise.
280         (BT_FN_LONG_PTR_LONG): Likewise.
281         (BT_FN_ULONG_PTR_ULONG): Likewise.
282         (BT_FN_VOID_UINT64_UINT64): Likewise.
283         (BT_FN_UINT8_UINT8_UINT8): Likewise.
284         (BT_FN_INT8_INT8_INT8): Likewise.
285         (BT_FN_INT16_INT16_INT16): Likewise.
286         (BT_FN_INT_INT_INT): Likewise.
287         (BT_FN_UINT_FLOAT_UINT): Likewise.
288         (BT_FN_FLOAT_UINT_UINT): Likewise.
289         (BT_FN_ULONG_UINT_UINT): Likewise.
290         (BT_FN_ULONG_UINT_PTR): Likewise.
291         (BT_FN_ULONG_ULONG_ULONG): Likewise.
292         (BT_FN_UINT_UINT_UINT): Likewise.
293         (BT_FN_VOID_UINT_PTR): Likewise.
294         (BT_FN_UINT_UINT_PTR: Likewise.
295         (BT_FN_UINT32_UINT64_PTR): Likewise.
296         (BT_FN_INT_INT_UINT_UINT): Likewise.
297         (BT_FN_UINT_UINT_UINT_UINT): Likewise.
298         (BT_FN_UINT_UINT_UINT_PTR): Likewise.
299         (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
300         (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
301         (BT_FN_LONG_LONG_UINT_UINT): Likewise.
302         (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
303         (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
304         (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
305         (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
306         (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
307         (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
308         * doc/frontends.texi: List BRIG FE.
309         * doc/install.texi (Testing): Add BRIG tesring requirements.
310         * doc/invoke.texi (Overall Options): Mention BRIG.
311         * doc/standards.texi (Standards): Doucment BRIG HSA version.
313 2017-01-24  Richard Biener  <rguenther@suse.de>
315         PR translation/79208
316         * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
318 2017-01-24  Martin Jambor  <mjambor@suse.cz>
320         PR bootstrap/79198
321         * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
322         * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
323         and known_contexts.
325 2017-01-24  Aldy Hernandez  <aldyh@redhat.com>
327         PR middle-end/79123
328         * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
329         casts from signed to unsigned really don't have a range.
331 2017-01-24  Markus Trippelsdorf  <markus@trippelsdorf.de>
333         * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
334         GMP_RNDx for compatiblity.
336 2017-01-24  Martin Liska  <mliska@suse.cz>
338         PR bootstrap/79132
339         * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
340         that would prevent us to call alloca with -1 as argument.
342 2017-01-24  Jakub Jelinek  <jakub@redhat.com>
344         * dwarf2out.c (output_compilation_unit_header, output_file_names):
345         Avoid -Wformat-security warning.
347 2017-01-23  Andrew Pinski  <apinski@cavium.com>
349         * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
350         cost table.
352 2017-01-23  Martin Sebor  <msebor@redhat.com>
354         PR middle-end/78703
355         * gimple-ssa-sprintf.c (warn_level): New global.
356         (format_integer): Use it here and throughout the rest of the file.
357         Use the same switch to compute sign as base.
358         (maybe_warn): New function.
359         (format_directive): Factor out warnings into maybe_warn.
360         Add debugging output.  Use warn_level.
361         (add_bytes): Use warn_level.
362         (pass_sprintf_length::compute_format_length): Add debugging output.
363         (try_substitute_return_value): Same.
364         (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
366         PR middle-end/78703
367         * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
368         (struct fmtresult, format_integer, format_floating): Adjust.
369         (fmtresult::fmtresult): Set max correctly in two argument ctor.
370         (get_string_length, format_string,format_directive): Same.
371         (pass_sprintf_length::compute_format_length): Same.
372         (try_substitute_return_value): Simplify slightly.
374         PR middle-end/78703
375         * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
376         (fmtresult::operator+=): Outlined.
377         (struct fmtresult): Add ctors.
378         (struct conversion_spec): Rename...
379         (struct directive): ...to this.  Add and remove data members.
380         (directive::set_width, directive::set_precision): New functions.
381         (format_percent): Use fmtresult ctor.
382         (get_width_and_precision): Remove.
383         (format_integer): Make naming changes.  Avoid computing width and
384         precision.
385         (format_floating): Same.  Adjust indentation.
386         (format_character, format_none): New functions.
387         (format_string): Moved character handling to format_character.
388         (format_directive): Remove arguments, change return type.
389         (parse_directive): New function.
390         (pass_sprintf_length::compute_format_length): Move directive
391         parsing to parse_directive.
393 2017-01-23  Jakub Jelinek  <jakub@redhat.com>
395         * tree.h (assign_assembler_name_if_neeeded): Rename to ...
396         (assign_assembler_name_if_needed): ... this.
397         * tree.c (assign_assembler_name_if_neeeded): Rename to ...
398         (assign_assembler_name_if_needed): ... this.
399         (free_lang_data_in_cgraph): Adjust callers.
400         * cgraphunit.c (cgraph_node::analyze): Likewise.
401         * omp-expand.c (expand_omp_taskreg, expand_omp_target):
402         Likewise.
404 2017-01-23  Richard Biener  <rguenther@suse.de>
406         PR tree-optimization/79088
407         PR tree-optimization/79188
408         * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
409         resetting loop bounds after last path deletion.  Reset loop
410         bounds of the target loop, make code match the comments.
411         * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
412         Make sure loops need no fixups.
414 2017-01-23  Kelvin Nilsen  <kelvin@gcc.gnu.org>
416         * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
417         exponent support with double type for first argument.
418         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
419         type returned by __builtin_vec_extract_sig,
420         __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
421         functions from "vector int" to "vector unsigned int" or from
422         "vector long long int" to "vector unsigned long long int".
423         Changed type returned by __builtin_vec_extract_exp,
424         __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
425         functions from "vector int" to "vector unsigned int" or from
426         "vector long long int" to "vector unsigned long long int".
427         Changed return type of __builtin_vec_test_data_class,
428         __builtin_vec_test_data_class_sp, and
429         __builtin_vec_test_data_class_dp from "vector int" to
430         "vector bool int" or from "vector long long int" to "vector bool
431         long long int" and changed second argument type from "unsigned
432         int" to "int".  Added new overloaded function forms "vector float
433         __builtin_vec_insert_exp (vector float, vector unsigned int)" and
434         "vector float __builtin_vec_insert_exp_sp (vector float, vector
435         unsigned int)" and "vector double __builtin_vec_insert_exp (vector
436         double, vector unsigned long long int)" and "vector double
437         __builtin_vec_insert_exp_dp (vector double, vector unsigned long
438         long int)".  Changed return type of
439         __builtin_scalar_test_data_class and
440         __builtin_scalar_test_data_class_sp and
441         __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
442         int" and changed second argument from "unsigned int" to "int".
443         Changed type returned by __builtin_scalar_test_neg,
444         __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
445         from "int" to "bool int".  Added new overloaded function form
446         "double __builtin_scalar_insert_exp (double, unsigned long long int)".
447         * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
448         exponent double-precision with floating point first argument.
449         * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
450         documentation of scalar_test_data_class, scalar_test_neg,
451         scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
452         vector_extract_exp, vec_extract_sig, vec_insert_exp, and
453         vec_test_data_class built-in functions to reflect refinements in
454         their type signatures.
456 2017-01-23  Andreas Tobler  <andreast@gcc.gnu.org>
458         * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
459         size of buf.
460         (aarch64_elf_asm_destructor): Likewise.
462 2017-01-23  Bernd Schmidt  <bschmidt@redhat.com>
464         PR rtl-optimization/78634
465         * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
466         (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
467         * ifcvt.c (noce_try_cmove): Add missing cost check.
469         PR rtl-optimization/71724
470         * combine.c (if_then_else_cond): Look for situations where it is
471         beneficial to undo the work of one of the recursive calls.
473 2017-01-23  Bin Cheng  <bin.cheng@arm.com>
475         PR tree-optimization/70754
476         * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
477         (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE.  Insert
478         combined stmt before it if not NULL.
479         (combine_chains): Process refs reversely and compute dominance point
480         for root ref.
482 2017-01-23  Martin Liska  <mliska@suse.cz>
484         PR tree-optimization/79196
485         * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
486         (fold_strstr_to_strncmp): ... this.  Fold the pattern to strncmp
487         instead of memcmp.
488         (strlen_optimize_stmt): Call the renamed function.
490 2017-01-23  Michael Matz  <matz@suse.de>
492         PR tree-optimization/78384
493         * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
495 2017-01-23  Richard Biener  <rguenther@suse.de>
497         PR tree-optimization/79186
498         * tree-vrp.c (register_new_assert_for): Make sure we've seen
499         both incoming edges before moving an assert.
501 2017-01-23  Martin Jambor  <mjambor@suse.cz>
503         * ipa-prop.c (load_from_param_1): Removed.
504         (load_from_unmodified_param): Bits from load_from_param_1 put back
505         here.
506         (load_from_param): Removed.
507         (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
508         with stmt.  Reverted back to use of load_from_unmodified_param.
510 2017-01-23  Martin Jambor  <mjambor@suse.cz>
512         PR ipa/79108
513         * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
514         (ipa_node_params): Annotate with GTY((for_user)).  Make descriptors
515         field a pointer to garbage collected vector, mark lattices and
516         ipcp_orig_node with GTY((skip)).
517         (ipa_get_param_count): Adjust to descriptors being a pointer.
518         (ipa_get_param): Likewise.
519         (ipa_get_type): Likewise.
520         (ipa_get_param_move_cost): Likewise.
521         (ipa_set_param_used): Likewise.
522         (ipa_get_controlled_uses): Likewise.
523         (ipa_set_controlled_uses): Likewise.
524         (ipa_is_param_used): Likewise.
525         (ipa_node_params_t): Move into garbage collector.  New methods insert
526         and remove.
527         (ipa_node_params_sum): Annotate wth GTY(()).
528         (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
529         garbage collected.
530         (ipa_load_from_parm_agg): Adjust declaration.
531         * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
532         * ipa-profile.c (ipa_profile): Likewise.
533         * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
534         (ipa_populate_param_decls): Make descriptors parameter garbage
535         collected.
536         (ipa_dump_param): Adjust to descriptors being a pointer.
537         (ipa_alloc_node_params): Likewise.
538         (ipa_initialize_node_params): Likewise.
539         (load_from_param_1): Make descriptors parameter garbage collected.
540         (load_from_unmodified_param): Likewise.
541         (load_from_param): Likewise.
542         (ipa_load_from_parm_agg): Likewise.
543         (ipa_node_params::~ipa_node_params): Removed.
544         (ipa_free_all_node_params): Remove call to delete operator.
545         (ipa_node_params_t::insert): New.
546         (ipa_node_params_t::remove): Likewise.
547         (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
548         copy known_csts and known_contexts vectors.
549         (ipa_read_node_info): Adjust to descriptors being a pointer.
550         (ipcp_modif_dom_walker): Make m_descriptors field garbage
551         collected.
552         (ipcp_transform_function): Make descriptors variable garbage
553         collected.
555 2017-01-23  Andrew Senkevich  <andrew.senkevich@intel.com>
557         * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
558         * config/i386/avx512dqintrin.h: Ditto.
559         * config/i386/avx512fintrin.h: Ditto.
560         * gcc/config/i386/i386.c: Handle new builtins.
561         * config/i386/i386-builtin.def: Add new builtins.
562         * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
563         (UNSPEC_KORTEST, UNSPEC_KTEST): New.
565 2017-01-23  Jakub Jelinek  <jakub@redhat.com>
566             Martin Liska  <mliska@suse.cz>
568         * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
569         * asan.c (asan_expand_poison_ifn): Support stores and use
570         appropriate ASAN report function.
571         * internal-fn.c (expand_ASAN_POISON_USE): New function.
572         * internal-fn.def (ASAN_POISON_USE): Declare.
573         * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
574         (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
575         * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
576         ASAN_POISON calls w/o LHS.
577         * tree-ssa.c (execute_update_addresses_taken): Create clobber
578         for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
579         from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
580         * gimplify.c (asan_poison_variables): Add attribute
581         use_after_scope_memory to variables that really needs to live
582         in memory.
583         * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
584         having the attribute.
586 2017-01-23  Martin Liska  <mliska@suse.cz>
588         * asan.c (create_asan_shadow_var): New function.
589         (asan_expand_poison_ifn): Likewise.
590         * asan.h (asan_expand_poison_ifn): New declaration.
591         * internal-fn.c (expand_ASAN_POISON): Likewise.
592         * internal-fn.def (ASAN_POISON): New builtin.
593         * sanopt.c (pass_sanopt::execute): Expand
594         asan_expand_poison_ifn.
595         * tree-inline.c (copy_decl_for_dup_finish): Make function
596         external.
597         * tree-inline.h (copy_decl_for_dup_finish): Likewise.
598         * tree-ssa.c (is_asan_mark_p): New function.
599         (execute_update_addresses_taken): Rewrite local variables
600         (identified just by use-after-scope as addressable) into SSA.
602 2017-01-22  Gerald Pfeifer  <gerald@pfeifer.com>
604         * doc/install.texi (Specific): opensource.apple.com uses https
605         now. Remove trailing slash.
607 2017-01-22  Gerald Pfeifer  <gerald@pfeifer.com>
609         * README.Portability: Remove note on an Irix compatibility issue.
611 2017-01-22  Dimitry Andric <dim@FreeBSD.org>
613         * gcov.c (INCLUDE_ALGORITHM): Define.
614         (INCLUDE_VECTOR): Define.
615         No longer include <vector> and <algorithm> directly.
617 2017-01-21  Gerald Pfeifer  <gerald@pfeifer.com>
619         * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
620         to https.
621         * doc/invoke.texi (Code Gen Options): Ditto.
623 2017-01-21  Jan Hubicka  <hubicka@ucw.cz>
625         PR lto/78407
626         * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
628 2017-01-21  Bernd Schmidt  <bschmidt@redhat.com>
630         rtl-optimization/79125
631         * cprop.c (local_cprop_pass): Handle cases where we make an
632         unconditional trap.
634 2017-01-20  Segher Boessenkool  <segher@kernel.crashing.org>
636         PR target/61729
637         PR target/77850
638         * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
639         read from, for big endian.
641 2017-01-20  Jiong Wang  <jiong.wang@arm.com>
643         * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
644         register pauth builtins for LP64 only.
646 2017-01-20  Marek Polacek  <polacek@redhat.com>
648         PR c/79152
649         * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
650         non-case labels.
652 2017-01-20  Alexander Monakov  <amonakov@ispras.ru>
654         * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
655         of safelen status.
656         * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
657         * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
658         * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
660 2017-01-20  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
662         PR target/71270
663         * config/arm/arm.c (neon_valid_immediate): Reject vector constants
664         in big-endian mode when they are not a single duplicated value.
666 2017-01-20  Richard Biener  <rguenther@suse.de>
668         * BASE-VER: Bump to 7.0.1.
670 2017-01-20  Alexander Monakov  <amonakov@ispras.ru>
672         * omp-low.c (omplow_simd_context): New struct.  Use it...
673         (lower_rec_simd_input_clauses): ...here and...
674         (lower_rec_input_clauses): ...here to hold common data.  Adjust all
675         references to idx, lane, max_vf, is_simt.
677 2017-01-20  Graham Markall  <graham.markall@embecosm.com>
679         * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
680         mcpu=nps400.
682 2017-01-20  Martin Jambor  <mjambor@suse.cz>
684         * hsa.h: Renaed to hsa-common.h.  Adjusted a comment.
685         * hsa.c: Renaed to hsa-common.c.  Change include of gt-hsa.h to
686         gt-hsa-common.h.
687         * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
688         (GTFILES): Rename hsa.c to hsa-common.c.
689         * hsa-brig.c: Change include of hsa.h to hsa-common.h.
690         * hsa-dump.c: Likewise.
691         * hsa-gen.c: Likewise.
692         * hsa-regalloc.c: Likewise.
693         * ipa-hsa.c: Likewise.
694         * omp-expand.c: Likewise.
695         * omp-low.c: Likewise.
696         * toplev.c: Likewise.
698 2017-01-20  Marek Polacek  <polacek@redhat.com>
700         PR c/64279
701         * doc/invoke.texi: Document -Wduplicated-branches.
702         * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
703         COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
704         POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
705         STATEMENT_LIST, and RETURN_EXPR.  For non-pure non-const functions
706         return 0 only when not OEP_LEXICOGRAPHIC.
707         (fold_build_cleanup_point_expr): Use the expression
708         location when building CLEANUP_POINT_EXPR.
709         * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
710         * tree.c (add_expr): Handle error_mark_node.
712 2017-01-20  Martin Liska  <mliska@suse.cz>
714         PR lto/69188
715         * tree-profile.c (init_ic_make_global_vars): Do not call
716         finalize_decl.
717         (gimple_init_gcov_profiler): Likewise.
719 2017-01-20  Martin Liska  <mliska@suse.cz>
721         PR ipa/71190
722         * cgraph.h (maybe_create_reference): Remove argument and
723         update comment.
724         * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
725         argument.
726         * ipa-cp.c (create_specialized_node): Likewise.
727         * symtab.c (symtab_node::maybe_create_reference): Handle
728         VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
730 2017-01-20  Martin Liska  <mliska@suse.cz>
732         * read-rtl-function.c (function_reader::create_function): Use
733         build_decl instread of build_decl_stat.
735 2017-01-20  Andrew Senkevich  <andrew.senkevich@intel.com>
737         * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
738         * config/i386/avx512dqintrin.h: Ditto.
739         * config/i386/avx512fintrin.h: Ditto.
740         * config/i386/i386-builtin-types.def: Add new types.
741         * gcc/config/i386/i386.c: Handle new types.
742         * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
743         (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
744         (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
745         (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
746         (__builtin_ia32_kshiftridi): New.
747         * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
749 2017-01-19  Segher Boessenkool  <segher@kernel.crashing.org>
751         PR target/78875
752         PR target/79140
753         * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
754         define to rs6000_init_stack_protect_guard.
755         (rs6000_init_stack_protect_guard): New function.
757 2017-01-19  Matthew Fortune  <matthew.fortune@imgtec.com>
758             Yunqiang Su  <yunqiang.su@imgtec.com>
760         * config.gcc (supported_defaults): Add madd4.
761         (with_madd4): Add validation.
762         (all_defaults): Add madd4.
763         * config/mips/mips.opt (mmadd4): New option.
764         * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
765         mmadd4.
766         (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
767         __mips_no_madd4.
768         (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
769         (ISA_HAS_FUSED_MADD4): Likewise.
770         * gcc/doc/invoke.texi (-mmadd4): Document the new option.
771         * gcc/doc/install.texi (--with-madd4): Document the new option.
773 2017-01-19  Jiong Wang  <jiong.wang@arm.com>
775         * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
776         entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
777         AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
778         (aarch64_init_pauth_hint_builtins): New.
779         (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
780         (aarch64_expand_builtin): Expand new builtins.
782 2017-01-19  Jiong Wang  <jiong.wang@arm.com>
784         * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
785         * combine-stack-adj.c (no_unhandled_cfa): Handle
786         REG_CFA_TOGGLE_RA_MANGLE.
787         * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
788         * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
789         info for return address signing.
790         (aarch64_expand_epilogue): Likewise.
792 2017-01-19  Jiong Wang  <jiong.wang@arm.com>
794         * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
795         * config/aarch64/aarch64-protos.h
796         (aarch64_return_address_signing_enabled): New declaration.
797         * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
798         New function.
799         (aarch64_expand_prologue): Sign return address before it's pushed onto
800         stack.
801         (aarch64_expand_epilogue): Authenticate return address fetched from
802         stack.
803         (aarch64_override_options): Sanity check for ILP32 and ISA level.
804         (aarch64_attributes): New function attributes for "sign-return-address".
805         * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
806         UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
807         ("*do_return"): Generate combined instructions according to key index.
808         ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
809         * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
810         iterators.
811         (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
812         * config/aarch64/aarch64.opt (msign-return-address=): New.
813         * doc/extend.texi (AArch64 Function Attributes): Documents
814         "sign-return-address=".
815         * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
817 2017-01-19  Matthew Fortune  <matthew.fortune@imgtec.com>
819         * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
820         overall option summary.
822 2017-01-19  Jiong Wang  <jiong.wang@arm.com>
824         * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
825         * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
826         AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
827         * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
829 2017-01-19  Michael Meissner  <meissner@linux.vnet.ibm.com>
831         * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
832         -mpower9-minmax by default for -mcpu=power9.
833         (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
834         128-bit floating point.
836 2017-01-20  Alan Modra  <amodra@gmail.com>
838         * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
839         optimizing for size.
841 2017-01-20  Alan Modra  <amodra@gmail.com>
843         PR target/79144
844         * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
845         for strcmp and strncmp from corresponding builtin decl.
847 2017-01-19  Uros Bizjak  <ubizjak@gmail.com>
849         * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
850         instead of i386/rtems-64.h.
851         * config/i386/rtems-64.h: Remove.
853 2017-01-19  Uros Bizjak  <ubizjak@gmail.com>
855         PR target/78478
856         Revert:
857         2013-11-05  Uros Bizjak  <ubizjak@gmail.com>
859         * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
861 2017-01-19  Tamar Christina  <tamar.christina@arm.com>
863         * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
864         Change int to HOST_WIDE_INT.
865         * config/aarch64/aarch64-protos.h
866         (aarch64_simd_gen_const_vector_dup): Likewise.
867         * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
869 2017-01-19  David Malcolm  <dmalcolm@redhat.com>
871         * langhooks-def.h (lhd_type_for_size): New decl.
872         (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
873         * langhooks.c (lhd_type_for_size): New function, taken from
874         lto_type_for_size.
876 2017-01-19  Pat Haugen  <pthaugen@us.ibm.com>
878         * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
879         define_bypass for CR latency.
880         (power9-cracked-alu): Update bypass latency and remove power9-branch.
881         (power9-alu2): Add define_bypass for CR latency.
882         (power9-cmp): New.
883         (power9-mul): Update insn latency.
884         (power9-mul-compare): Update insn latency, bypass latency and remove
885         power9-branch.
887 2016-01-19  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
889         * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
890         Delete.
891         * config/aarch64/aarch64.md
892         (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
893         aarch64_nopcrelative_literal_loads.
894         (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
896 2017-01-19  Chenghua Xu  <paul.hua.gm@gmail.com>
898         * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
899         TARGET_LOONGSON_3A.
900         (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
902 2017-01-19  Doug Gilmore  <doug.gilmore@imgtec.com>
904         PR target/78176
905         * config.gcc (supported_defaults): Add lxc1-sxc1.
906         (with_lxc1_sxc1): Add validation.
907         (all_defaults): Add lxc1-sxc1.
908         * config/mips/mips.opt (mlxc1-sxc1): New option.
909         * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
910         mlxc1-sxc1.
911         (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
912         __mips_no_lxc1_sxc1.
913         (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
914         * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
915         * doc/install.texi (--with-lxc1-sxc1): Document the new option.
917 2017-01-19  Richard Biener  <rguenther@suse.de>
919         PR tree-optimization/72488
920         * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
921         sure to restore SSA info.
922         * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
924 2017-01-19  Richard Earnshaw  <rearnsha@arm.com>
926         PR rtl-optimization/79121
927         * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
928         of the inner type when shifting an extended value.
930 2017-01-17  Jan Hubicka  <hubicka@ucw.cz>
932         PR lto/78407
933         * symtab.c (symtab_node::equal_address_to): Fix comparing of
934         interposable aliases.
936 2017-01-18  Peter Bergner  <bergner@vnet.ibm.com>
938         PR target/78516
939         * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
940         Use the evmergelohi instruction.
941         (mov_si<mode>_e500_subreg4_2_le): Likewise.
942         (mov_sitf_e500_subreg8_2_be): Likewise.
943         (mov_sitf_e500_subreg12_2_le): Likewise.
944         (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
945         (mov_si<mode>_e500_subreg4_2_be): Likewise.
946         (mov_sitf_e500_subreg8_2_le): Likewise.
947         (mov_sitf_e500_subreg12_2_be): Likewise.
949 2017-01-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
951         * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
952         attribute from vecsimple to vecperm.
953         (altivec_vbpermq2): Likewise.
955 2017-01-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
957         PR target/79040
958         * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
960 2017-01-18  Aaron Sawdey  <acsawdey@linux.vnet.ibm.com>
961         * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
962         * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
963         strcmp. Fix bug where comparison didn't stop with zero byte. Fix
964         case where N arg is SIZE_MAX.
965         * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
966         (cmpstrsi): Add pattern.
968 2017-01-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
970         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
971         __builtin_vec_revb builtins.
972         * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
973         built-in functions to support generation of the ISA 3.0 XXBR<x>
974         vector byte reverse instructions.
975         (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
976         (P9V_BUILTIN_XXBRD_V2DI): Likewise.
977         (P9V_BUILTIN_XXBRD_V2DF): Likewise.
978         (P9V_BUILTIN_XXBGW_V4SI): Likewise.
979         (P9V_BUILTIN_XXBGW_V4SF): Likewise.
980         (P9V_BUILTIN_XXBGH_V8HI): Likewise.
981         (P9V_BUILTIN_VEC_REVB): Likewise.
982         * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
983         generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
984         (p9_xxbrq_v16qi): Likewise.
985         (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
986         (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
987         (p9_xxbrh_v8hi): Likewise.
988         * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
989         * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
990         vec_revb built-in functions.
992 2017-01-18  Uros Bizjak  <ubizjak@gmail.com>
994         PR rtl-optimization/78952
995         * config/i386/i386.md (any_extract): New code iterator.
996         (*insvqi_2): Use any_extract for source operand.
997         (*insvqi_3): Use any_shiftrt for source operand.
999 2017-01-18  Wilco Dijkstra  <wdijkstr@arm.com>
1001         * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
1002         New function.
1003         (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
1005 2017-01-18  Matthias Klose  <doko@ubuntu.com>
1007         * doc/install.texi: Allow default for --with-target-bdw-gc-include.
1009 2016-01-18  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1011         * config/rs6000/altivec.h (vec_bperm): Change #define.
1012         * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
1013         (altivec_vbpermq2): New define_insn.
1014         (altivec_vbpermd): Likewise.
1015         * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
1016         function interface.
1017         (VBPERMD): Likewise.
1018         (VBPERM): New polymorphic function interface.
1019         * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
1020         Add entries for P9V_BUILTIN_VEC_VBPERM.
1021         * doc/extend.texi: Add interfaces for vec_bperm.
1023 2017-01-18  Andreas Krebbel  <krebbel@linux.vnet.ibm.com>
1025         * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
1026         first letter of error messages.
1027         (s390_resolve_overloaded_builtin): Likewise.
1028         * config/s390/s390.c (s390_expand_builtin): Likewise.
1029         (s390_invalid_arg_for_unprototyped_fn): Likewise.
1030         (s390_valid_target_attribute_inner_p): Likewise.
1031         * config/s390/s390.md ("tabort"): Likewise.
1033 2017-01-18  Toma Tabacu  <toma.tabacu@imgtec.com>
1035         * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
1036         (ISA_AVOID_DIV_HILO): New macro.
1037         (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
1038         (ISA_HAS_DDIV): Likewise.
1040 2017-01-18  Markus Trippelsdorf  <markus@trippelsdorf.de>
1042         * doc/invoke.texi (fabi-version): Correct number of occurrences.
1044 2017-01-18  Markus Trippelsdorf  <markus@trippelsdorf.de>
1046         * doc/invoke.texi (fabi-version): Spelling fix.
1048 2017-01-18  Markus Trippelsdorf  <markus@trippelsdorf.de>
1050         PR c++/70182
1051         * doc/invoke.texi (fabi-version): Mention mangling fix for
1052         operator names.
1054 2017-01-18  Markus Trippelsdorf  <markus@trippelsdorf.de>
1056         PR c++/77489
1057         * doc/invoke.texi (fabi-version): Document discriminator mangling.
1059 2017-01-17  Segher Boessenkool  <segher@kernel.crashing.org>
1061         PR target/78875
1062         * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
1063         * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
1064         the new options.
1065         * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
1066         flexible settings.
1067         (stack_protect_test): Ditto.
1068         * config/rs6000/rs6000.opt (mstack-protector-guard=,
1069         mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
1070         options.
1071         * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
1072         Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
1073         -mstack-protector-guard-offset=.
1074         (RS/6000 and PowerPC Options): Ditto.
1076 2017-01-17  Uros Bizjak  <ubizjak@gmail.com>
1078         * config/i386/i386.h (MASK_CLASS_P): New define.
1079         * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
1080         there are no registers from different register sets also when
1081         mask registers are used.  Update function comment.
1082         * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
1083         to (*k/*r) and (*k/*km) alternatives.
1085 2017-01-17  Wilco Dijkstra  <wdijkstr@arm.com>
1087         * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
1088         * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
1089         (EH_RETURN_HANDLER_RTX): New define.
1090         * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
1091         Force frame pointer in EH return functions.
1092         (aarch64_expand_epilogue): Add barrier for eh_return.
1093         (aarch64_final_eh_return_addr): Remove.
1094         (aarch64_eh_return_handler_rtx): New function.
1095         * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
1096         Remove.
1097         (aarch64_eh_return_handler_rtx): New prototype.
1099 2017-01-17  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1101         * config/rs6000/altivec.h (vec_rlmi): New #define.
1102         (vec_vrlnm): Likewise.
1103         (vec_rlnm): Likewise.
1104         * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
1105         (UNSPEC_VRLNM): Likewise.
1106         (VIlong): New mode iterator.
1107         (altivec_vrl<VI_char>mi): New define_insn.
1108         (altivec_vrl<VI_char>nm): Likewise.
1109         * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
1110         function entry.
1111         (VRLDNM): Likewise.
1112         (RLNM): New polymorphic function entry.
1113         (VRLWMI): New monomorphic function entry.
1114         (VRLDMI): Likewise.
1115         (RLMI): New polymorphic function entry.
1116         * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
1117         new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
1118         * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
1119         vec_vrlnm.
1121 2017-01-17  Jakub Jelinek  <jakub@redhat.com>
1123         PR debug/78839
1124         * dwarf2out.c (field_byte_offset): Restore the
1125         PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
1126         and DECL_FIELD_BIT_OFFSET.  Use fold_build2 instead of build2 + fold.
1127         (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
1128         of build2 + fold.
1130 2017-01-17  Eric Botcazou  <ebotcazou@adacore.com>
1132         PR ada/67205
1133         * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
1135 2017-01-17  Jakub Jelinek  <jakub@redhat.com>
1137         PR debug/71669
1138         * dwarf2out.c (add_data_member_location_attribute): For constant
1139         offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
1140         instead of DW_AT_data_member_location, DW_AT_bit_offset and
1141         DW_AT_byte_size attributes.
1143 2017-01-17  Eric Botcazou  <ebotcazou@adacore.com>
1145         * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
1146         after forcing to constant memory when the code model is medium.
1148 2017-01-17  Julia Koval  <julia.koval@intel.com>
1150         PR target/76731
1151         * config/i386/avx512fintrin.h
1152         (_mm512_i32gather_ps): Change __addr type to void const*.
1153         (_mm512_mask_i32gather_ps): Ditto.
1154         (_mm512_i32gather_pd): Ditto.
1155         (_mm512_mask_i32gather_pd): Ditto.
1156         (_mm512_i64gather_ps): Ditto.
1157         (_mm512_mask_i64gather_ps): Ditto.
1158         (_mm512_i64gather_pd): Ditto.
1159         (_mm512_mask_i64gather_pd): Ditto.
1160         (_mm512_i32gather_epi32): Ditto.
1161         (_mm512_mask_i32gather_epi32): Ditto.
1162         (_mm512_i32gather_epi64): Ditto.
1163         (_mm512_mask_i32gather_epi64): Ditto.
1164         (_mm512_i64gather_epi32): Ditto.
1165         (_mm512_mask_i64gather_epi32): Ditto.
1166         (_mm512_i64gather_epi64): Ditto.
1167         (_mm512_mask_i64gather_epi64): Ditto.
1168         (_mm512_i32scatter_ps): Change __addr type to void*.
1169         (_mm512_mask_i32scatter_ps): Ditto.
1170         (_mm512_i32scatter_pd): Ditto.
1171         (_mm512_mask_i32scatter_pd): Ditto.
1172         (_mm512_i64scatter_ps): Ditto.
1173         (_mm512_mask_i64scatter_ps): Ditto.
1174         (_mm512_i64scatter_pd): Ditto.
1175         (_mm512_mask_i64scatter_pd): Ditto.
1176         (_mm512_i32scatter_epi32): Ditto.
1177         (_mm512_mask_i32scatter_epi32): Ditto.
1178         (_mm512_i32scatter_epi64): Ditto.
1179         (_mm512_mask_i32scatter_epi64): Ditto.
1180         (_mm512_i64scatter_epi32): Ditto.
1181         (_mm512_mask_i64scatter_epi32): Ditto.
1182         (_mm512_i64scatter_epi64): Ditto.
1183         (_mm512_mask_i64scatter_epi64): Ditto.
1184         * config/i386/avx512pfintrin.h
1185         (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
1186         (_mm512_mask_prefetch_i32gather_ps): Ditto.
1187         (_mm512_mask_prefetch_i64gather_pd): Ditto.
1188         (_mm512_mask_prefetch_i64gather_ps): Ditto.
1189         (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
1190         (_mm512_prefetch_i32scatter_ps): Ditto.
1191         (_mm512_mask_prefetch_i32scatter_pd): Ditto.
1192         (_mm512_mask_prefetch_i32scatter_ps): Ditto.
1193         (_mm512_prefetch_i64scatter_pd): Ditto.
1194         (_mm512_prefetch_i64scatter_ps): Ditto.
1195         (_mm512_mask_prefetch_i64scatter_pd): Ditto.
1196         (_mm512_mask_prefetch_i64scatter_ps): Ditto.
1197         * config/i386/avx512vlintrin.h
1198         (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
1199         (_mm_mmask_i32gather_ps): Ditto.
1200         (_mm256_mmask_i32gather_pd): Ditto.
1201         (_mm_mmask_i32gather_pd): Ditto.
1202         (_mm256_mmask_i64gather_ps): Ditto.
1203         (_mm_mmask_i64gather_ps): Ditto.
1204         (_mm256_mmask_i64gather_pd): Ditto.
1205         (_mm_mmask_i64gather_pd): Ditto.
1206         (_mm256_mmask_i32gather_epi32): Ditto.
1207         (_mm_mmask_i32gather_epi32): Ditto.
1208         (_mm256_mmask_i32gather_epi64): Ditto.
1209         (_mm_mmask_i32gather_epi64): Ditto.
1210         (_mm256_mmask_i64gather_epi32): Ditto.
1211         (_mm_mmask_i64gather_epi32): Ditto.
1212         (_mm256_mmask_i64gather_epi64): Ditto.
1213         (_mm_mmask_i64gather_epi64): Ditto.
1214         (_mm256_i32scatter_ps): Change __addr type to void*.
1215         (_mm256_mask_i32scatter_ps): Ditto.
1216         (_mm_i32scatter_ps): Ditto.
1217         (_mm_mask_i32scatter_ps): Ditto.
1218         (_mm256_i32scatter_pd): Ditto.
1219         (_mm256_mask_i32scatter_pd): Ditto.
1220         (_mm_i32scatter_pd): Ditto.
1221         (_mm_mask_i32scatter_pd): Ditto.
1222         (_mm256_i64scatter_ps): Ditto.
1223         (_mm256_mask_i64scatter_ps): Ditto.
1224         (_mm_i64scatter_ps): Ditto.
1225         (_mm_mask_i64scatter_ps): Ditto.
1226         (_mm256_i64scatter_pd): Ditto.
1227         (_mm256_mask_i64scatter_pd): Ditto.
1228         (_mm_i64scatter_pd): Ditto.
1229         (_mm_mask_i64scatter_pd): Ditto.
1230         (_mm256_i32scatter_epi32): Ditto.
1231         (_mm256_mask_i32scatter_epi32): Ditto.
1232         (_mm_i32scatter_epi32): Ditto.
1233         (_mm_mask_i32scatter_epi32): Ditto.
1234         (_mm256_i32scatter_epi64): Ditto.
1235         (_mm256_mask_i32scatter_epi64): Ditto.
1236         (_mm_i32scatter_epi64): Ditto.
1237         (_mm_mask_i32scatter_epi64): Ditto.
1238         (_mm256_i64scatter_epi32): Ditto.
1239         (_mm256_mask_i64scatter_epi32): Ditto.
1240         (_mm_i64scatter_epi32): Ditto.
1241         (_mm_mask_i64scatter_epi32): Ditto.
1242         (_mm256_i64scatter_epi64): Ditto.
1243         (_mm256_mask_i64scatter_epi64): Ditto.
1244         (_mm_i64scatter_epi64): Ditto.
1245         (_mm_mask_i64scatter_epi64): Ditto.
1246         * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
1247         (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
1248         (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
1249         (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
1250         (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
1251         (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
1252         (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
1253         (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
1254         (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
1255         (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
1256         (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
1257         (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
1258         (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
1259         (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
1260         (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
1261         (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
1262         (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
1263         (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
1264         (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
1265         (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
1266         (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
1267         (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
1268         (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
1269         (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
1270         (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
1271         (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
1272         (VOID_QI_V8DI_PCINT_INT_INT): Remove.
1273         (V16SF_V16SF_PCVOID_V16SI_HI_INT,  V8DF_V8DF_PCVOID_V8SI_QI_INT)
1274         (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
1275         (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
1276         (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
1277         (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
1278         (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
1279         (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
1280         (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
1281         (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
1282         (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
1283         (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
1284         (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
1285         (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
1286         (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
1287         (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
1288         (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
1289         (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
1290         (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
1291         (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
1292         (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
1293         (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
1294         (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
1295         (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
1296         (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
1297         (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
1298         (VOID_QI_V8DI_PCVOID_INT_INT): Add.
1299         * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
1300         definitions accordingly.
1302 2017-01-17  Kito Cheng  <kito.cheng@gmail.com>
1303             Kuan-Lin Chen  <kuanlinchentw@gmail.com>
1305         PR target/79079
1306         * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
1307         gen_lowpart.
1309 2017-01-17  Vladimir Makarov  <vmakarov@redhat.com>
1311         PR target/79058
1312         * ira-conflicts.c (ira_build_conflicts): Update total conflict
1313         hard regs for inner regno.
1315 2017-01-17  Martin Liska  <mliska@suse.cz>
1317         PR ipa/71207
1318         * ipa-polymorphic-call.c (contains_type_p): Fix wrong
1319         assumption and add comment.
1321 2017-01-17  Nathan Sidwell  <nathan@acm.org>
1323         * ipa-visibility.c (localize_node): New function, broken out of ...
1324         (function_and_variable_visibility): ... here. Call it.
1326 2017-01-17  Jan Hubicka  <hubicka@ucw.cz>
1328         PR middle-end/77445
1329         * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
1330         correctly set frequency of oudgoing edge.
1331         (duplicate_thread_path): Fix profile updating.
1333 2017-01-17  Jakub Jelinek  <jakub@redhat.com>
1335         PR other/79046
1336         * configure.ac: Add GCC_BASE_VER.
1337         * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
1338         version from BASE-VER file.
1339         (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
1340         (gcc.o): Depend on $(BASEVER).
1341         * common.opt (dumpfullversion): New option.
1342         * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
1343         * doc/invoke.texi: Document -dumpfullversion.
1344         * doc/install.texi: Document --with-gcc-major-version-only.
1345         * configure: Regenerated.
1347 2017-01-17  Richard Biener  <rguenther@suse.de>
1349         PR tree-optimization/71433
1350         * tree-vrp.c (register_new_assert_for): Merge same asserts
1351         on all incoming edges.
1352         (process_assert_insertions_for): Handle insertions at the
1353         beginning of BBs.
1355 2017-01-17  Gerald Pfeifer  <gerald@pfeifer.com>
1357         * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
1358         * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
1360 2017-01-17  Kaz Kojima  <kkojima@gcc.gnu.org>
1362         PR target/78633
1363         * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
1364         RTL sharing.
1366 2017-01-17  Alan Modra  <amodra@gmail.com>
1368         PR target/79066
1369         * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
1370         * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
1371         symbolic stack limit when pic.
1373 2017-01-16  Martin Sebor  <msebor@redhat.com>
1375         PR tree-optimization/78608
1376         * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
1378 2017-01-16  Jeff Law  <law@redhat.com>
1380         Revert:
1381         2016-12-02  Tadek Kijkowski  <tkijkowski@gmail.com>
1382         * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
1383         for several include directories that may be relative to sysroot.
1384         * config/i386/x-mingw32 (gplus_includedir): Define.
1385         (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
1386         (native_system_includedir): Likewise.
1387         * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
1388         override if TARGET_SYSTEM_ROOT is defined.
1389         (NATIVE_SYSTEM_HEADER_DIR): Likewise.
1391         PR tree-optimization/79090
1392         PR tree-optimization/33562
1393         PR tree-optimization/61912
1394         PR tree-optimization/77485
1395         * tree-ssa-dse.c (compute_trims): Accept STMT argument.  Dump STMT
1396         and computed trims into the dump file.
1398 2017-01-17  Uros Bizjak  <ubizjak@gmail.com>
1400         * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
1402 2017-01-16  Jakub Jelinek  <jakub@redhat.com>
1404         PR c/79089
1405         * gimplify.c (gimplify_init_constructor): If want_value and
1406         object == lhs, unshare lhs to avoid invalid tree sharing.  Formatting
1407         fix.
1409         PR target/79080
1410         * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
1411         sequence.  Formatting fixes.
1412         (doloop_optimize): Formatting fixes.
1414         PR driver/49726
1415         * gcc.c (debug_level_greater_than_spec_func): New function.
1416         (static_spec_functions): Add debug-level-gt spec function.
1417         (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
1418         !g0.
1419         * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
1420         * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
1421         * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
1422         gpubnames, ggnu-pubnames, gno-record-gcc-switches,
1423         grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
1424         gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
1426 2017-01-16  Uros Bizjak  <ubizjak@gmail.com>
1428         * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
1429         QImode fixups to general and mask registers only.
1431 2017-01-16  Carl Love  <cel@us.ibm.com>
1433         * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
1434         for built-in functions
1435         vector signed char vec_nabs (vector signed char)
1436         vector signed short vec_nabs (vector signed short)
1437         vector signed int vec_nabs (vector signed int)
1438         vector signed long long vec_nabs (vector signed long long)
1439         vector float vec_nabs (vector float)
1440         vector double vec_nabs (vector double)
1441         * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
1442         and NABS overload.
1443         * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
1444         * config/rs6000/altivec.h: New define for vec_nabs built-in function.
1445         * doc/extend.texi: Update the documentation file for the new built-in
1446         functions.
1448 2017-01-16  Martin Sebor  <msebor@redhat.com>
1450         * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
1451         message.
1453 2017-01-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1455         * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
1456         UNSPEC_VSX__XXSPLTD to require special splat handling.
1458 2017-01-16  David Malcolm  <dmalcolm@redhat.com>
1460         PR bootstrap/78616
1461         * system.h: Poison strndup.
1463 2017-01-16  Alan Modra  <amodra@gmail.com>
1465         PR target/79098
1466         * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
1467         use a switch.
1469 2017-01-16  Georg-Johann Lay  <avr@gjlay.de>
1471         * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
1473 2017-01-15  Uros Bizjak  <ubizjak@gmail.com>
1475         * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
1476         call recog here.  Assert that INSN_CODE (insn) is non-negative.
1478 2017-01-15  Segher Boessenkool  <segher@kernel.crashing.org>
1480         PR target/72749
1481         * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
1482         fallthrough.
1483         * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
1484         in the currently scheduled RTL fragment.
1486 2017-01-15  Segher Boessenkool  <segher@kernel.crashing.org>
1488         PR rtl-optimization/78751
1489         * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
1490         give up.
1492 2017-01-14  Jeff Law  <law@redhat.com>
1494         PR tree-optimization/79090
1495         * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
1496         variable length stores.
1497         (compute_trims): Delete dead assignment to *trim_tail.
1498         (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
1499         zero length.
1501 2017-01-14  Bernd Schmidt  <bschmidt@redhat.com>
1503         PR rtl-optimization/78626
1504         PR rtl-optimization/78727
1505         * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
1506         of a block, and split such blocks after everything else is finished.
1508 2017-01-14  Alan Modra  <amodra@gmail.com>
1510         PR target/72749
1511         * combine.c (recog_for_combine_1): Set INSN_CODE before calling
1512         target legitimate_combined_insn.
1513         * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
1514         (rs6000_legitimate_combined_insn): New function.
1515         * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
1516         all uses.
1517         (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
1518         (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
1519         (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
1521 2017-01-14  Gerald Pfeifer  <gerald@pfeifer.com>
1523         * doc/frontends.texi (G++ and GCC): Remove references to Java.
1525 2017-01-13  Jeff Law  <law@redhat.com>
1527         PR tree-optimization/33562
1528         PR tree-optimization/61912
1529         PR tree-optimization/77485
1530         * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
1531         a statement.
1532         (delete_dead_assignment): Likewise.
1533         (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
1534         statement to delete_dead_call and delete_dead_assignment.
1536 2017-01-13  David Malcolm  <dmalcolm@redhat.com>
1538         PR c/78304
1539         * substring-locations.c (format_warning_va): Strengthen case 1 so
1540         that both endpoints of the substring must be within the format
1541         range for just the substring to be printed.
1543 2017-01-13  Uros Bizjak  <ubizjak@gmail.com>
1545         * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
1546         * config/i386/i386.c (ix86_target_string): Add missing options
1547         to isa_opts and reorder options by implied ISAs.  Rename isa_opts2 to
1548         isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
1549         flags_other and ix86_target_other to flags2_other.  Display unknown
1550         isa2 options.
1551         (ix86_valid_target_attribute_inner_p): Add missing options and
1552         reorder options by implied ISAs, as in ix86_target_string.
1554 2017-01-13  Richard Sandiford  <richard.sandiford@arm.com>
1556         * hash-table.h (hash_table::too_empty_p): New function.
1557         (hash_table::expand): Use it.
1558         (hash_table::traverse): Likewise.
1559         (hash_table::empty_slot): Use sizeof (value_type) instead of
1560         sizeof (PTR) to convert bytes to elements.  Shrink the table
1561         if the current size is excessive for the current number of
1562         elements.
1564 2017-01-13  Richard Sandiford  <richard.sandiford@arm.com>
1566         * ira-costs.c (record_reg_classes): Break from the inner loop
1567         early once alt_fail is known to be true.  Update outer loop
1568         handling accordingly.
1570 2017-01-13  Jeff Law  <law@redhat.com>
1572         * tree-ssa-dse.c (decrement_count): New function.
1573         (increment_start_addr, maybe_trim_memstar_call): Likewise.
1574         (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
1575         when we know the partially dead statement is a mem* function.
1577         PR tree-optimization/61912
1578         PR tree-optimization/77485
1579         * tree-ssa-dse.c: Include expr.h.
1580         (maybe_trim_constructor_store): New function.
1581         (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
1583         PR tree-optimization/33562
1584         PR tree-optimization/61912
1585         PR tree-optimization/77485
1586         * doc/invoke.texi: Document new dse-max-object-size param.
1587         * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
1588         * tree-ssa-dse.c: Include params.h.
1589         (dse_store_status): New enum.
1590         (initialize_ao_ref_for_dse): New, partially extracted from
1591         dse_optimize_stmt.
1592         (valid_ao_ref_for_dse, normalize_ref): New.
1593         (setup_live_bytes_from_ref, compute_trims): Likewise.
1594         (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
1595         (maybe_trim_partially_dead_store): Likewise.
1596         (maybe_trim_complex_store): Likewise.
1597         (dse_classify_store): Renamed from dse_possibly_dead_store_p.
1598         Track what bytes live from the original store.  Return tri-state
1599         for dead, partially dead or live.
1600         (dse_dom_walker): Add constructor, destructor and new private members.
1601         (delete_dead_call, delete_dead_assignment): New extracted from
1602         dse_optimize_stmt.
1603         (dse_optimize_stmt): Make a member of dse_dom_walker.
1604         Use initialize_ao_ref_for_dse.
1606         PR tree-optimization/33562
1607         PR tree-optimization/61912
1608         PR tree-optimization/77485
1609         * sbitmap.h (bitmap_count_bits): Prototype.
1610         (bitmap_clear_range, bitmap_set_range): Likewise.
1611         * sbitmap.c (bitmap_clear_range): New function.
1612         (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
1614 2017-01-13  Martin Liska  <mliska@suse.cz>
1616         PR ipa/79043
1617         * function.c (set_cfun): Add new argument force.
1618         * function.h (set_cfun): Likewise.
1619         * ipa-inline-transform.c (inline_call): Use the function when
1620         strict alising from is dropped for function we inline to.
1622 2017-01-13  Richard Biener  <rguenther@suse.de>
1624         * tree-pretty-print.c (dump_generic_node): Fix inverted condition
1625         for dumping GIMPLE INTEGER_CSTs.
1627 2017-01-13  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
1629         * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
1630         to 201112L since C++17.
1632 2017-01-13  Maxim Ostapenko  <m.ostapenko@samsung.com>
1634         PR sanitizer/78887
1635         * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
1636         if -fsanitize=kernel-address is present.
1638 2017-01-13  Richard Biener  <rguenther@suse.de>
1640         * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
1641         as _Literal ( type ) number in case usual suffixes do not
1642         preserve all information.
1644 2017-01-13  Richard Biener  <rguenther@suse.de>
1646         PR tree-optimization/77283
1647         * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
1648         and ssa-iterators.h.
1649         (is_feasible_trace): Implement a cost model based on joiner
1650         PHI node uses.
1652 2017-01-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
1654         PR target/79004
1655         * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
1656         char or short to __float128/_Float128 directly.
1658 2017-01-12  Martin Sebor  <msebor@redhat.com>
1660         to -Wformat-overflow.
1661         * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
1662         (min_bytes_remaining): Same.
1663         (get_string_length): Same.
1664         (format_string): Same.
1665         (format_directive): Same.
1666         (add_bytes): Same.
1667         (pass_sprintf_length::handle_gimple_call): Same.
1669 2017-01-12  Jakub Jelinek  <jakub@redhat.com>
1671         * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
1672         info.nowrite calls with no lhs that can't throw.  Return bool
1673         whether gsi_remove has been called or not.
1674         (pass_sprintf_length::handle_gimple_call): Return bool whether
1675         try_substitute_return_value called gsi_remove.  Formatting fix.
1676         (pass_sprintf_length::execute): Don't use gsi_remove if
1677         handle_gimple_call returned true.
1679         PR bootstrap/79069
1680         * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
1681         be removed due to side-effects, don't remove following barrier nor
1682         turn the successor edge into fallthru edge.
1684 2017-01-12  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1686         PR target/79044
1687         * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
1688         element-reversing loads and stores as not swappable.
1690 2017-01-12  Nathan Sidwell  <nathan@acm.org>
1691             Nicolai Stange  <nicstange@gmail.com>
1693         * combine.c (try_combine): Don't ignore result of overlap checking
1694         loop.  Combine overlap & asm check into single loop.
1696 2017-01-12  Richard Biener  <rguenther@suse.de>
1698         * tree-pretty-print.c (dump_generic_node): Provide -gimple
1699         variant for MEM_REF.  Sanitize INTEGER_CST for -gimple.
1701 2017-01-12  Richard Biener  <rguenther@suse.de>
1703         * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
1704         and TS_TARGET_OPTION directly derive from TS_BASE.
1705         * tree-core.h (tree_optimization_option): Derive from tree_base.
1706         (tree_target_option): Likewise.
1708 2017-01-11  Uros Bizjak  <ubizjak@gmail.com>
1710         * config/i386/i386.c (memory_address_length): Increase len
1711         only when rip_relative_addr_p returns false.
1713 2017-01-11  Julia Koval  <julia.koval@intel.com>
1715         * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
1716         (OPTION_MASK_ISA_SGX_SET): New.
1717         (ix86_handle_option): Handle OPT_msgx.
1718         * config.gcc: Added sgxintrin.h.
1719         * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
1720         * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
1721         * config/i386/i386.c (ix86_target_string): Add -msgx.
1722         (PTA_SGX): New.
1723         (ix86_option_override_internal): Handle new options.
1724         (ix86_valid_target_attribute_inner_p): Add sgx.
1725         * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
1726         * config/i386/i386.opt: Add msgx.
1727         * config/i386/sgxintrin.h: New file.
1728         * config/i386/x86intrin.h: Add sgxintrin.h.
1730 2017-01-11  Jakub Jelinek  <jakub@redhat.com>
1732         PR c++/71537
1733         * fold-const.c (maybe_nonzero_address): Return 1 for function
1734         local objects.
1735         (tree_single_nonzero_warnv_p): Don't handle function local objects
1736         here.
1738         PR c++/72813
1739         * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
1740         of c-header.
1742 2017-01-11  David Malcolm  <dmalcolm@redhat.com>
1744         PR driver/78877
1745         * opts.c: Include "spellcheck.h"
1746         (struct string_fragment): New struct.
1747         (struct edit_distance_traits<const string_fragment &>): New
1748         struct.
1749         (get_closest_sanitizer_option): New function.
1750         (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
1752 2017-01-11  Jakub Jelinek  <jakub@redhat.com>
1754         * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
1755         by 12.
1756         (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
1757         DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
1758         (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
1759         (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
1760         for initial die_offset if dwarf_split_debug_info.
1761         (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
1762         initial next_die_offset if dwo_id is non-NULL.  Don't emit padding
1763         fields.
1764         (output_skeleton_debug_sections): Formatting fix.  Use
1765         DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
1766         DWARF_COMPILE_UNIT_HEADER_SIZE.  Don't emit padding.
1768 2017-01-11  Wilco Dijkstra  <wdijkstr@arm.com>
1770         * config/arm/cortex-a53.md: Add bypasses for
1771         cortex_a53_r2f_cvt.
1772         (cortex_a53_r2f): Only use for transfers.
1773         (cortex_a53_f2r): Likewise.
1774         (cortex_a53_r2f_cvt): Add reservation for conversions.
1775         (cortex_a53_f2r_cvt): Likewise.
1777 2017-01-11  Tamar Christina  <tamar.christina@arm.com>
1779         * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
1780         to all inlined functions, change static to extern.
1782 2017-01-11  Christophe Lyon  <christophe.lyon@linaro.org>
1784         PR target/78253
1785         * config/arm/arm.c (legitimize_pic_address): Handle reference to
1786         weak symbol.
1787         (arm_assemble_integer): Likewise.
1789 2017-01-11  Richard Earnshaw  <rearnsha@arm.com>
1791         * config.gcc: Use new awk script to check CPU, FPU and architecture
1792         parameters for --with-... options.
1793         * config/arm/parsecpu.awk: New file
1794         * config/arm/arm-cpus.in: New file.
1795         * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
1796         files.
1797         * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
1798         files.
1799         * config/arm/t-arm: Update dependency rules.
1800         * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
1801         of processing .def files.
1802         * config/arm/genopt.sh: Deleted.
1803         * config/arm/gentune.sh: Deleted.
1804         * config/arm/arm-cores.def: Deleted.
1805         * config/arm/arm-arches.def: Deleted.
1806         * config/arm/arm-fpus.def: Deleted.
1807         * config/arm/arm-tune.md: Regenerated.
1808         * config/arm/arm-tables.opt: Regenerated.
1809         * config/arm/arm-cpu.h: New generated file.
1810         * config/arm/arm-cpu-data.h: New generated file.
1811         * config/arm/arm-cpu-cdata.h: New generated file.
1813 2017-01-11  Maxim Ostapenko  <m.ostapenko@samsung.com>
1815         PR lto/79042
1816         * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
1817         bit.
1818         (input_varpool_node): Unpack dynamically_initialized bit.
1820 2017-01-11  Eric Botcazou  <ebotcazou@adacore.com>
1822         PR rtl-optimization/79032
1823         * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
1824         the alignment of the adjusted memory reference against that of MODE,
1825         instead of the alignment of the original memory reference.
1827 2017-01-11  Martin Jambor  <mjambor@suse.cz>
1829         * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
1830         test.
1831         * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
1832         decorated functions.
1834 2017-01-11  Richard Biener  <rguenther@suse.de>
1836         * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
1837         set range/nonnull info for PHI results.  Do not set it on
1838         stmts marked for removal.
1840 2017-01-10  Eric Botcazou  <ebotcazou@adacore.com>
1842         * expr.c (store_field): In the bitfield case, fetch the return value
1843         from the registers before applying a single big-endian adjustment.
1844         Always do a final load for a BLKmode value not larger than a word.
1846 2017-01-10  David Malcolm  <dmalcolm@redhat.com>
1848         PR c++/77949
1849         * input.c (selftest::test_accessing_ordinary_linemaps): Verify
1850         that we correctly handle column numbers greater than
1851         LINE_MAP_MAX_COLUMN_NUMBER.
1853 2017-01-10  Martin Sebor  <msebor@redhat.com>
1855         PR middle-end/78245
1856         * gimple-ssa-sprintf.c (get_destination_size): Call
1857         {init,fini}object_sizes.
1858         * tree-object-size.c (addr_object_size): Adjust.
1859         (pass_through_call): Adjust.
1860         (pass_object_sizes::execute): Adjust.
1861         * tree-object-size.h (fini_object_sizes): Declare.
1863 2017-01-10  Martin Sebor  <msebor@redhat.com>
1865         PR tree-optimization/78775
1866         * builtins.c (get_size_range): Move...
1867         * calls.c: ...to here.
1868         (alloc_max_size): Accept zero argument.
1869         (operand_signed_p): Remove.
1870         (maybe_warn_alloc_args_overflow): Call get_size_range.
1871         * calls.h (get_size_range): Declare.
1873 2017-01-10  Joe Seymour  <joe.s@somniumtech.com>
1875         * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
1876         from TI's devices.csv file as of September 2016.
1877         * config/msp430/msp430.c (msp430_mcu_data): Likewise.
1879 2017-01-10  Sandra Loosemore  <sandra@codesourcery.com>
1881         * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
1882         * doc/invoke.texi: Likewise.
1883         * doc/md.texi: Likewise.
1884         * doc/objc.texi: Likewise.
1886 2017-01-10  Joshua Conner  <joshconner@google.com>
1888         * config/arm/fuchsia-elf.h: New file.
1889         * config/fuchsia.h: New file.
1890         * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
1891         (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
1892         targets.
1893         * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
1895 2016-01-10  Richard Biener  <rguenther@suse.de>
1897         PR tree-optimization/79034
1898         * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
1899         Propagate out degenerate PHIs in the joiner.
1901 2017-01-10  Martin Liska  <mliska@suse.cz>
1903         * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
1904         (sort_congruence_classes_by_decl_uid): Likewise.
1905         (sort_congruence_class_groups_by_decl_uid): Likewise.
1906         (sem_item_optimizer::merge_classes): Sort class, groups in these
1907         classes and members in the groups by DECL_UID of declarations.
1908         This would make merge operations stable.
1910 2017-01-10  Martin Liska  <mliska@suse.cz>
1912         * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
1913         usage of m_classes_vec.
1914         (sem_item_optimizer::~sem_item_optimizer):  Likewise.
1915         (sem_item_optimizer::get_group_by_hash): Likewise.
1916         (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
1917         (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
1918         (sem_item_optimizer::verify_classes): Likewise.
1919         (sem_item_optimizer::process_cong_reduction): Likewise.
1920         (sem_item_optimizer::dump_cong_classes): Likewise.
1921         (sem_item_optimizer::merge_classes): Likewise.
1922         * ipa-icf.h (congruence_class_hash): Rename from
1923         congruence_class_group_hash.  Remove declaration of m_classes_vec.
1925 2017-01-10  Andrew Senkevich  <andrew.senkevich@intel.com>
1927         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
1928         OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
1929         * config.gcc: Add avx512vpopcntdqintrin.h.
1930         * config/i386/avx512vpopcntdqintrin.h: New.
1931         * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
1932         * config/i386/i386-builtin-types.def: Add new types.
1933         * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
1934         __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
1935         __builtin_ia32_vpopcountq_v8di_mask): New.
1936         * config/i386/i386-c.c (ix86_target_macros_internal): Define
1937         __AVX512VPOPCNTDQ__.
1938         * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
1939         (PTA_AVX512VPOPCNTDQ): Define.
1940         * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
1941         TARGET_AVX512VPOPCNTDQ_P): Define.
1942         * config/i386/i386.opt: Add mavx512vpopcntdq.
1943         * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
1944         * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
1946 2017-01-01  Jan Hubicka  <hubicka@ucw.cz>
1948         PR middle-end/77484
1949         * predict.def (PRED_CALL): Set to 67.
1951 2017-01-09  Eric Botcazou  <ebotcazou@adacore.com>
1953         * expr.c (store_field): In the bitfield case, if the value comes from
1954         a function call and is of an aggregate type returned in registers, do
1955         not modify the field mode; extract the value in all cases if the mode
1956         is BLKmode and the size is not larger than a word.
1958 2017-01-09  Dominique d'Humieres  <dominiq@lps.ens.fr>
1960         PR target/71017
1961         * config/i386/cpuid.h: Fix undefined behavior.
1963 2017-01-04  Jeff Law  <law@redhat.com>
1965         PR tree-optimization/79007
1966         PR tree-optimization/67955
1967         * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
1968         conservative for pt.null when flag_non_call_exceptions is on.
1970 2017-01-09  Jakub Jelinek  <jakub@redhat.com>
1972         PR translation/79019
1973         PR translation/79020
1974         * params.def (PARAM_INLINE_MIN_SPEEDUP,
1975         PARAM_IPA_CP_SINGLE_CALL_PENALTY,
1976         PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
1977         in descriptions.
1978         * config/avr/avr.opt (maccumulate-args): Likewise.
1979         * config/msp430/msp430.opt (mwarn-mcu): Likewise.
1980         * common.opt (freport-bug): Likewise.
1981         * cif-code.def (CIF_FINAL_ERROR): Likewise.
1982         * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
1983         * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
1984         translatable string.
1985         * config/i386/i386.c (function_value_32): Likewise.
1986         * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
1987         * config/msp430/msp430.c (msp430_option_override, msp430_attr):
1988         Likewise.
1989         * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
1990         * common/config/msp430/msp430-common.c (msp430_handle_option):
1991         Likewise.
1992         * symtab.c (symtab_node::verify_base): Likewise.
1993         * opts.c (set_debug_level): Likewise.
1994         * tree.c (verify_type_variant): Likewise.  Fix typo in comment.
1995         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
1996         missing whitespace to translatable strings.
1997         * config/avr/avr.md (bswapsi2): Fix typo in comment.
1998         * config/sh/superh.h: Likewise.
1999         * config/i386/xopintrin.h: Likewise.
2000         * config/i386/znver1.md: Likewise.
2001         * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
2002         * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
2003         * double-int.h (struct double_int): Likewise.
2004         * double-int.c (div_and_round_double): Likewise.
2005         * wide-int.cc: Likewise.
2006         * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
2007         * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
2008         * cfgcleanup.c (crossjumps_occured): Renamed to ...
2009         (crossjumps_occurred): ... this.
2010         (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
2011         Adjust all uses.
2013         PR tree-optimization/78899
2014         * tree-if-conv.c (version_loop_for_if_conversion): Instead of
2015         returning bool return struct loop *, NULL for failure and the new
2016         loop on success.
2017         (versionable_outer_loop_p): Don't version outer loop if it has
2018         dont_vectorized bit set.
2019         (tree_if_conversion): When versioning outer loop, ensure
2020         tree_if_conversion is performed also on the inner loop of the
2021         non-vectorizable outer loop copy.
2022         * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix.  Fold
2023         LOOP_VECTORIZED in inner loop of the scalar outer loop and
2024         prevent vectorization of it.
2025         (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
2026         the outer loop vectorization of the non-scalar version is attempted
2027         before vectorization of the inner loop in scalar version.  If
2028         outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
2029         vectorization of its inner loop.
2030         * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
2031         has 2 inner loops, rename also on edges from bb whose single pred
2032         is outer_loop->header.  Fix typo in function comment.
2034 2017-01-09  Martin Sebor  <msebor@redhat.com>
2036         PR bootstrap/79033
2037         * asan.c (asan_emit_stack_protection): Increase local buffer size
2038         to avoid snprintf truncation warning.
2040 2017-01-09  Andrew Pinski  <apinski@cavium.com>
2042         * config/aarch64/aarch64-cores.def: Add thunderx2t99.  Change vulcan
2043         to reference thunderx2t99 for the tuning structure
2044         * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
2045         Rename to ...
2046         (thunderx2t99_extra_costs): This.
2047         * config/aarch64/aarch64-tune.md: Regenerate.
2048         * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
2049         (vulcan_addrcost_table): This.
2050         (vulcan_regmove_cost): Rename to ...
2051         (thunderx2t99_regmove_cost): This.
2052         (vulcan_vector_cost): Rename to ...
2053         (thunderx2t99_vector_cost): this.
2054         (vulcan_branch_cost): Rename to ...
2055         (thunderx2t99_branch_cost): This.
2056         (vulcan_tunings): Rename to ...
2057         (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
2058         * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
2060 2017-01-09  Martin Jambor  <mjambor@suse.cz>
2062         PR ipa/78365
2063         PR ipa/78599
2064         * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
2065         * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
2066         (propagate_vr_accross_jump_function): Use the above function for all
2067         value range computations for pass-through jump functions and type
2068         converasion from explicit value range values.
2069         (ipcp_propagate_stage): Do not attempt to deduce types of formal
2070         parameters from TYPE_ARG_TYPES.
2071         * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
2072         (ipa_write_node_info): Stream type of the actual argument.
2073         (ipa_read_node_info): Likewise. Also remove trailing whitespace.
2075 2017-01-09  Martin Liska  <mliska@suse.cz>
2077         PR pch/78970
2078         * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
2079         (lookup_compiler): Do not show error message with have_E.
2081 2017-01-09  Jakub Jelinek  <jakub@redhat.com>
2083         PR tree-optimization/78938
2084         * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
2085         where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
2086         BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
2087         {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR.  Formatting
2088         fixes.
2090 2017-01-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2092         * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
2093         is const0_rtx.
2095 2017-01-09  Richard Biener  <rguenther@suse.de>
2097         PR tree-optimization/78997
2098         * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
2099         name condition properly.
2101 2017-01-09  Richard Biener  <rguenther@suse.de>
2103         PR debug/79000
2104         * dwarf2out.c (is_cxx): New overload with context.
2105         (is_naming_typedef_decl): Use it.
2107 2017-01-08  Sandra Loosemore  <sandra@codesourcery.com>
2109         * invoke.texi (Option Summary): Correct spacing in option lists
2110         and add line breaks to fix over-long lines.
2112 2017-01-08  Sandra Loosemore  <sandra@codesourcery.com>
2114         PR middle-end/17660
2116         * extend.texi (Common Variable Attributes): Add xref to GCC
2117         Internals manual to explain mode attribute keywords.
2119 2017-01-08  Sandra Loosemore  <sandra@codesourcery.com>
2121         PR other/16519
2122         * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
2123         and Preprocessor Options.
2124         (Options for Linking): Document -pthread here....
2125         (RS/6000 and PowerPC Options): ...not here.
2126         (Solaris 2 Options): ...or here.
2127         * doc/cppopts.texi: Document -pthread.
2129 2017-01-08  Martin Sebor  <msebor@redhat.com>
2131         PR middle-end/77708
2132         * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
2133         * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
2134         New member functions.
2135         (format_directive): Used them.
2136         (add_bytes): Same.
2137         (pass_sprintf_length::handle_gimple_call): Same.
2138         * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
2139         to avoid truncation for any argument.
2140         (extract_affine_mul): Same.
2141         * tree.c (get_file_function_name): Same.
2143 2017-01-01  Jan Hubicka  <hubicka@ucw.cz>
2145         PR middle-end/77484
2146         * predict.def (PRED_INDIR_CALL): Set to 86.
2148 2017-01-07  Sandra Loosemore  <sandra@codesourcery.com>
2150         PR preprocessor/54124
2151         * doc/cppopts.texi: Reformat -d subtable to list the full name
2152         of the options.  Add cross-reference to the docs for the general
2153         compiler -d options.
2154         * doc/invoke.texi (Developer Options): Add cross-reference to the
2155         preprocessor-specific -d option documentation.
2157 2017-01-07  Sandra Loosemore  <sandra@codesourcery.com>
2159         PR preprocessor/13498
2160         * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
2161         redudant material, and reflect new command-line options.
2162         (System Headers): Likewise.
2164 2017-01-07  Sandra Loosemore  <sandra@codesourcery.com>
2166         * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
2167         -isystem, and -idirafter.  Copy-edit.
2168         * doc/cppopts.texi: Copy-edit.  Remove contradiction about
2169         default for -ftrack-macro-expansion.  Delete obsolete and
2170         badly-formatted implementation details about -fdebug-cpp output.
2171         * doc/cppwarnopts.texi: Copy-edit.
2173 2017-01-07  David Malcolm  <dmalcolm@redhat.com>
2175         PR c++/72803
2176         * input.c (selftest::test_accessing_ordinary_linemaps): Verify
2177         that the transition from a max line width >= 1<<10 to narrower
2178         lines works correctly.
2180 2017-01-07  Alexandre Oliva <aoliva@redhat.com>
2182         * doc/options.texi (PerFunction): New.
2183         * opt-functions.awk (switch_flags): Map both Optimization and
2184         PerFunction to CL_OPTIMIZATION.
2185         * opth-gen.awk: Test for PerFunction flag along with
2186         Optimization.
2187         * optc-save-gen.awk: Likewise.  Introduce var_opt_hash and set
2188         it only when the latter is present.  Skip those that don't in
2189         the hash function generator.
2190         * common.opt (fvar-tracking): Mark as PerFunction instead of
2191         Optimization.
2192         (fvar-tracking-assignments): Likewise.
2193         (fvar-tracking-assignments-toggle): Likewise.
2194         (fvar-tracking-uninit): Likewise.
2196 2017-01-07  Jakub Jelinek  <jakub@redhat.com>
2198         PR translation/79018
2199         * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
2200         the and store.
2202 2017-01-06  Mikael Pettersson  <mikpelinux@gmail.com>
2204         PR target/57583
2205         * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
2206         * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
2207         TARGET_LONG_JUMP_TABLE_OFFSETS.
2208         * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
2209         * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
2210         * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
2211         (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
2212         * config/m68k/m68k.md (tablejump expander): Likewise.
2213         (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
2214         TARGET_LONG_JUMP_TABLE_OFFSETS.
2215         (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
2216         * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
2218 2017-01-06  Edgar E. Iglesias <edgar.iglesias@xilinx.com>
2219             David Holsgrove <david.holsgrove@xilinx.com>
2221         * common/config/microblaze/microblaze-common.c
2222         (TARGET_EXCEPT_UNWIND_INFO): Remove.
2223         * config/microblaze/microblaze-protos.h (microblaze_eh_return):
2224         New prototype.
2225         * config/microblaze/microblaze.c (microblaze_must_save_register)
2226         (microblaze_expand_epilogue, microblaze_return_addr): Handle
2227         calls_eh_return.
2228         (microblaze_eh_return): New function.
2229         * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
2230         (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
2231         (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
2232         * config/microblaze/microblaze.md (eh_return): New pattern.
2234 2017-01-06  Jakub Jelinek  <jakub@redhat.com>
2236         * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
2237         GCC_DIAGNOSTIC_STRINGIFY): Define.
2239         * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
2241 2017-01-06  Andre Vieira  <andre.simoesdiasvieira@arm.com>
2243         * config/arm/arm.md (<mcrr>): New.
2244         (<mrrc>): New.
2245         * config/arm/arm.c (arm_arch5te): New.
2246         (arm_option_override): Set arm_arch5te.
2247         (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
2248         and mrrc2.
2249         * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
2250         (arm_mcrr_qualifiers): ... this. New.
2251         (MRRC_QUALIFIERS): Define to...
2252         (arm_mrrc_qualifiers): ... this. New.
2253         * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
2254         __arm_mrrc2): New.
2255         * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
2256         * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
2257         (MRRCI, mrrc, MRRC): New.
2258         * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
2259         VUNSPEC_MRRC2): New.
2261 2017-01-06  Andre Vieira  <andre.simoesdiasvieira@arm.com>
2263         * config/arm/arm.md (<mcr>): New.
2264         (<mrc>): New.
2265         * config/arm/arm.c (arm_coproc_builtin_available): Add
2266         support for mcr, mrc, mcr2 and mrc2.
2267         * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
2268         (arm_mcr_qualifiers): ... this. New.
2269         (MRC_QUALIFIERS): Define to ...
2270         (arm_mrc_qualifiers): ... this. New.
2271         (MCR_QUALIFIERS): Define to ...
2272         (arm_mcr_qualifiers): ... this. New.
2273         * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
2274         __arm_mrc2): New.
2275         * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
2276         * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
2277         * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
2278         VUNSPEC_MRC2): New.
2280 2017-01-06  Andre Vieira  <andre.simoesdiasvieira@arm.com>
2282         * config/arm/arm.md (*ldc): New.
2283         (*stc): New.
2284         (<ldc>): New.
2285         (<stc>): New.
2286         * config/arm/arm.c (arm_coproc_builtin_available): Add
2287         support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
2288         (arm_coproc_ldc_stc_legitimate_address): New.
2289         * config/arm/arm-builtins.c (arm_type_qualifiers): Add
2290         'qualifier_const_pointer'.
2291         (LDC_QUALIFIERS): Define to...
2292         (arm_ldc_qualifiers): ... this. New.
2293         (STC_QUALIFIERS): Define to...
2294         (arm_stc_qualifiers): ... this. New.
2295         * config/arm/arm-protos.h
2296         (arm_coproc_ldc_stc_legitimate_address): New.
2297         * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
2298         __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
2299         * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
2300         stc2, stcl, stc2l): New.
2301         * config/arm/constraints.md (Uz): New.
2302         * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
2303         * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
2304         VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
2305         VUNSPEC_STC2L): New.
2307 2017-01-06  Andre Vieira  <andre.simoesdiasvieira@arm.com>
2309         * config/arm/arm.md (<cdp>): New.
2310         * config/arm/arm.c (neon_const_bounds): Rename this ...
2311         (arm_const_bounds): ... this.
2312         (arm_coproc_builtin_available): New.
2313         * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
2314         (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
2315         (CDP_QUALIFIERS): Define to...
2316         (arm_cdp_qualifiers): ... this. New.
2317         (void_UP): Define.
2318         (arm_expand_builtin_args): Add case for 6 arguments.
2319         * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
2320         (arm_const_bounds): ... this.
2321         (arm_coproc_builtin_available): New.
2322         * config/arm/arm_acle.h (__arm_cdp): New.
2323         (__arm_cdp2): New.
2324         * config/arm/arm_acle_builtins.def (cdp): New.
2325         (cdp2): New.
2326         * config/arm/iterators.md (CDPI,CDP,cdp): New.
2327         * config/arm/neon.md: Rename all 'neon_const_bounds' to
2328         'arm_const_bounds'.
2329         * config/arm/types.md (coproc): New.
2330         * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
2331         * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
2332         * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
2333         arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
2335 2017-01-06  Andre Vieira  <andre.simoesdiasvieira@arm.com>
2337         * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
2338         (UBINOP_QUALIFIERS): New.
2339         (si_UP): Define.
2340         (acle_builtin_data): New. Change comment.
2341         (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
2342         ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
2343         ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
2344         arm_acle_builtins.def.
2345         (ARM_BUILTIN_ACLE_PATTERN_START): Define.
2346         (arm_init_acle_builtins): New.
2347         (CRC32_BUILTIN): Remove.
2348         (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
2349         crc32cb, crc32ch and crc32cw.
2350         (arm_init_crc32_builtins): Remove.
2351         (arm_init_builtins): Use arm_init_acle_builtins rather
2352         than arm_init_crc32_builtins.
2353         (arm_expand_acle_builtin): New.
2354         (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
2355         * config/arm/arm_acle_builtins.def: New.
2357 2017-01-06  Andre Vieira  <andre.simoesdiasvieira@arm.com>
2359         * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
2360         (arm_builtin_datum): ... this.
2361         (arm_init_neon_builtin): Rename to ...
2362         (arm_init_builtin): ... this. Add a new parameters PREFIX
2363         and USE_SIG_IN_NAME.
2364         (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
2365         'arm_init_builtin'. Replace type 'neon_builtin_datum' with
2366         'arm_builtin_datum'.
2367         (arm_init_vfp_builtins): Likewise.
2368         (builtin_arg): Rename enum's replacing 'NEON_ARG' with
2369         'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
2370         (arm_expand_neon_args): Rename to ...
2371         (arm_expand_builtin_args): ... this. Rename builtin_arg
2372         enum values and differentiate between ARG_BUILTIN_MEMORY
2373         and ARG_BUILTIN_NEON_MEMORY.
2374         (arm_expand_neon_builtin_1): Rename to ...
2375         (arm_expand_builtin_1): ... this. Rename builtin_arg enum
2376         values, arm_expand_builtin_args and add bool parameter NEON.
2377         (arm_expand_neon_builtin): Use arm_expand_builtin_1.
2378         (arm_expand_vfp_builtin): Likewise.
2379         (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
2381 2017-01-01  Jan Hubicka  <hubicka@ucw.cz>
2383         PR middle-end/77484
2384         * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
2385         * predict.c (tree_estimate_probability_bb): Reverse direction of
2386         polymorphic call predictor.
2388 2017-01-06  David Malcolm  <dmalcolm@redhat.com>
2390         * passes.c (execute_one_pass): Split out pass-skipping logic into...
2391         (determine_pass_name_match): ...this new function and...
2392         (should_skip_pass_p): ...this new function.
2394 2017-01-06  Nathan Sidwell  <nathan@acm.org>
2396         * ipa-visibility.c (function_and_variable_visibility): Reformat
2397         comments and long lines.  Remove extrneous if.
2398         * symtab.c (symtab_node::make_decl_local): Fix code format.
2399         (symtab_node::set_section_for_node): Fix comment typo.
2401 2017-01-06  Martin Liska  <mliska@suse.cz>
2403         PR bootstrap/79003
2404         * lra-constraints.c: Rename invariant to lra_invariant.
2405         * predict.c (set_even_probabilities): Initialize e to NULL.
2407 2017-01-05  Martin Sebor  <msebor@redhat.com>
2409         PR tree-optimization/78910
2410         * gimple-ssa-sprintf.c (tree_digits): Add an argument.
2411         (format_integer): Correct off-by-one error in the handling
2412         of precision with negative numbers in signed conversions..
2414 2017-01-05  Eric Botcazou  <ebotcazou@adacore.com>
2416         * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
2418 2017-01-05  Jakub Jelinek  <jakub@redhat.com>
2420         PR tree-optimization/71016
2421         * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
2422         factor_out_conditional_conversion.  Formatting fix.
2423         (factor_out_conditional_conversion): Add cond_stmt argument.
2424         If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
2425         cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
2426         Formatting fix.
2428 2017-01-05  David Malcolm  <dmalcolm@redhat.com>
2430         * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
2431         read-rtl-function.o, and selftest-rtl.o.
2432         * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
2433         (selftest::aarch64_test_loading_full_dump): New function.
2434         (selftest::aarch64_run_selftests): New function.
2435         (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
2436         selftest::aarch64_run_selftests.
2437         * config/i386/i386.c
2438         (selftest::ix86_test_loading_dump_fragment_1): New function.
2439         (selftest::ix86_test_loading_call_insn): New function.
2440         (selftest::ix86_test_loading_full_dump): New function.
2441         (selftest::ix86_test_loading_unspec): New function.
2442         (selftest::ix86_run_selftests): Call the new functions.
2443         * emit-rtl.c (maybe_set_max_label_num): New function.
2444         * emit-rtl.h (maybe_set_max_label_num): New decl.
2445         * function.c (instantiate_decls): Guard call to
2446         instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
2447         * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
2448         "static".
2449         * gensupport.c (gen_reader::gen_reader): Pass "false"
2450         for new "compact" param of rtx_reader.
2451         * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
2452         rather than an empty string for NULL strings.
2453         * read-md.c: Potentially include config.h rather than bconfig.h.
2454         Wrap include of errors.h with #ifdef GENERATOR_FILE.
2455         (have_error): New global, copied from errors.c.
2456         (md_reader::read_name): Rename to...
2457         (md_reader::read_name_1): ...this, adding "out_loc" param,
2458         and converting "missing name or number" to returning false, rather
2459         than failing.
2460         (md_reader::read_name): Reimplement in terms of read_name_1.
2461         (md_reader::read_name_or_nil): New function.
2462         (md_reader::read_string): Handle "(nil)" by returning NULL.
2463         (md_reader::md_reader): Add new param "compact".
2464         (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
2465         (md_reader::read_file): New method.
2466         * read-md.h (md_reader::md_reader): Add new param "compact".
2467         (md_reader::read_file): New method.
2468         (md_reader::is_compact): New accessor.
2469         (md_reader::read_name): Convert return type from void to file_location.
2470         (md_reader::read_name_or_nil): New decl.
2471         (md_reader::read_name_1): New decl.
2472         (md_reader::m_compact): New field.
2473         (noop_reader::noop_reader): Pass "false" for new "compact" param
2474         of rtx_reader.
2475         (rtx_reader::rtx_reader): Add new "compact" param.
2476         (rtx_reader::read_rtx_operand): Make virtual and convert return
2477         type from void to rtx.
2478         (rtx_reader::read_until): New decl.
2479         (rtx_reader::handle_any_trailing_information): New virtual function.
2480         (rtx_reader::postprocess): New virtual function.
2481         (rtx_reader::finalize_string): New virtual function.
2482         (rtx_reader::m_in_call_function_usage): New field.
2483         (rtx_reader::m_reuse_rtx_by_id): New field.
2484         * read-rtl-function.c: New file.
2485         * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
2486         * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
2487         (selftest::verify_three_block_rtl_cfg): New decl.
2488         * read-rtl-function.h: New file.
2489         * read-rtl.c: Potentially include config.h rather than bconfig.h.
2490         For host, include function.h, memmodel.h, and emit-rtl.h.
2491         (one_time_initialization): New function.
2492         (struct compact_insn_name): New struct.
2493         (compact_insn_names): New array.
2494         (find_code): Handle insn codes in compact dumps.
2495         (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
2496         (bind_subst_iter_and_attr): Likewise.
2497         (add_condition_to_string): Likewise.
2498         (add_condition_to_rtx): Likewise.
2499         (apply_attribute_uses): Likewise.
2500         (add_current_iterators): Likewise.
2501         (apply_iterators): Likewise.
2502         (initialize_iterators): Guard usage of apply_subst_iterator with
2503         #ifdef GENERATOR_FILE.
2504         (read_conditions): Wrap with #ifdef GENERATOR_FILE.
2505         (md_reader::read_mapping): Likewise.
2506         (add_define_attr_for_define_subst): Likewise.
2507         (add_define_subst_attr): Likewise.
2508         (read_subst_mapping): Likewise.
2509         (check_code_iterator): Likewise.
2510         (rtx_reader::read_rtx): Likewise.  Move one-time initialization
2511         logic to...
2512         (one_time_initialization): New function.
2513         (rtx_reader::read_until): New method.
2514         (read_flags): New function.
2515         (parse_reg_note_name): New function.
2516         (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
2517         Handle reuse_rtx ids.
2518         Wrap iterator lookup within #ifdef GENERATOR_FILE.
2519         Add parsing support for RTL dumps, mirroring the special-cases in
2520         print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
2521         values, and calling handle_any_trailing_information.
2522         (rtx_reader::read_rtx_operand): Convert return type from void
2523         to rtx, returning return_rtx.  Handle case 'e'.  Call
2524         finalize_string on XSTR and XTMPL fields.
2525         (rtx_reader::read_nested_rtx):  Handle dumps in which trailing
2526          "(nil)" values were omitted.  Call the postprocess vfunc on the
2527         return_rtx.
2528         (rtx_reader::rtx_reader): Add new "compact" param and pass to base
2529         class ctor.  Initialize m_in_call_function_usage.  Call
2530         one_time_initialization.
2531         * rtl-tests.c (selftest::test_uncond_jump): Call
2532         set_new_first_and_last_insn.
2533         * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
2534         * selftest-rtl.c: New file.
2535         * selftest-rtl.h (class selftest::rtl_dump_test): New class.
2536         (selftest::get_insn_by_uid): New decl.
2537         * selftest-run-tests.c (selftest::run_tests): Call
2538         read_rtl_function_c_tests.
2539         * selftest.h  (selftest::read_rtl_function_c_tests): New decl.
2540         * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
2541         dumps.
2543 2017-01-05  Uros Bizjak  <ubizjak@gmail.com>
2545         * config/i386/i386.md (*testqi_ext_3): No need to handle memory
2546         operands in a special way.  Assert that pos+len <= mode precision.
2548 2017-01-05  Jakub Jelinek  <jakub@redhat.com>
2550         * common.opt (fvect-cost-model): Remove RejectNegative flag, use
2551         3 argument Alias with unlimited for the negative form.
2552         (fno-vect-cost-model): Removed.
2554 2017-01-05  Martin Liska  <mliska@suse.cz>
2556         * hsa-gen.c (gen_hsa_divmod): New function.
2557         (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
2559 2017-01-05  Martin Liska  <mliska@suse.cz>
2561         PR pch/78970
2562         * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
2563         header.
2565 2017-01-05  Andreas Krebbel  <krebbel@linux.vnet.ibm.com>
2567         * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
2568         small constant length operands.
2570 2017-01-05  Andreas Krebbel  <krebbel@linux.vnet.ibm.com>
2572         * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
2573         between loop iterations.
2575 2017-01-05  Martin Liska  <mliska@suse.cz>
2577         PR sanitizer/78815
2578         * gimplify.c (gimplify_decl_expr): Compare to
2579         asan_poisoned_variables instread of checking flags.
2580         (gimplify_target_expr): Likewise.
2581         (gimplify_expr): Likewise.
2582         (gimplify_function_tree): Conditionally initialize
2583         asan_poisoned_variables.
2585 2017-01-04  Jeff Law  <law@redhat.com>
2587         PR tree-optimizatin/78812
2588         * rtl.h (contains_mem_rtx_p): Prototype.
2589         * ifcvt.c (containts_mem_rtx_p): Move from here to...
2590         * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
2591         * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
2592         and prune MEMs that are not at the toplevel of a SET_SRC rtx.  Look
2593         through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
2595 2017-01-04  Alexandre Oliva <aoliva@redhat.com>
2597         * input.c (assert_char_at_range): Default-initialize actual_range.
2599 2017-01-04  Alexandre Oliva <aoliva@redhat.com>
2601         * df-scan.c (df_ref_create_structure): Make regno unsigned,
2602         to match the caller.
2604 2017-01-04  Alexandre Oliva <aoliva@redhat.com>
2606         * cfgexpand.c (expand_gimple_basic_block): Disregard debug
2607         insns after final jump in test to emit dummy move.
2609 2017-01-04  Alexandre Oliva <aoliva@redhat.com>
2611         * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
2612         * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
2614 2017-01-04  Alexandre Oliva <aoliva@redhat.com>
2616         * multiple_target.c (create_dispatcher_calls): Init e_next.
2617         * tree-ssa-loop-split.c (split_loop): Init border.
2618         * tree-vect-loop.c (vect_determine_vectorization_factor): Init
2619         scalar_type.
2621 2017-01-04  Michael Meissner  <meissner@linux.vnet.ibm.com>
2623         PR target/71977
2624         PR target/70568
2625         PR target/78823
2626         * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
2627         (altivec_register_operand): Do not return true if the operand
2628         contains a SUBREG mixing SImode and SFmode.
2629         (vsx_register_operand): Likewise.
2630         (vsx_reg_sfsubreg_ok): New predicate.
2631         (vfloat_operand): Do not return true if the operand contains a
2632         SUBREG mixing SImode and SFmode.
2633         (vint_operand): Likewise.
2634         (vlogical_operand): Likewise.
2635         (gpc_reg_operand): Likewise.
2636         (int_reg_operand): Likewise.
2637         * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
2638         * config/rs6000/rs6000.c (valid_sf_si_move): New function to
2639         determine if a MOVSI or MOVSF operation contains SUBREGs that mix
2640         SImode and SFmode.
2641         (rs6000_emit_move_si_sf_subreg): New helper function.
2642         (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
2643         fixup SUBREGs involving SImode and SFmode.
2644         * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
2645         numbers for the new peephole2 optimization.
2646         (peephole2 for SFmode unions): New peephole2 to optimize cases in
2647         the GLIBC math library that do AND/IOR/XOR operations on single
2648         precision floating point.
2649         * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
2650         target macros to say whether we need to avoid SUBREGs mixing
2651         SImode and SFmode.
2652         (TARGET_ALLOW_SF_SUBREG): Likewise.
2653         * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
2654         (UNSPEC_SI_FROM_SF): Likewise.
2655         (iorxor): Change spacing.
2656         (and_ior_xor): New iterator for AND, IOR, and XOR.
2657         (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
2658         (movdi_from_sf_zero_ext): Likewise.
2659         (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
2660         instead of gpc_reg_operand.  Add SImode/SFmode SUBREG support.
2661         (movsf_from_si): New insn for SImode/SFmode SUBREG support.
2662         (fma<mode>4): Use gpc_reg_operand instead of register_operand.
2663         (fms<mode>4): Likewise.
2664         (fnma<mode>4): Likewise.
2665         (fnms<mode>4): Likewise.
2666         (nfma<mode>4): Likewise.
2667         (nfms<mode>4): Likewise.
2669 2017-01-04  Marek Polacek  <polacek@redhat.com>
2671         PR c++/64767
2672         * doc/invoke.texi: Document -Wpointer-compare.
2674 2017-01-04  Jakub Jelinek  <jakub@redhat.com>
2676         * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
2677         RejectNegative.
2679         * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
2680         descriptions for -gdwarf-5 and emit them as uleb128 instead of
2681         2-byte data.
2683 2017-01-04  Kelvin Nilsen  <kelvin@gcc.gnu.org>
2685         PR target/78056
2686         * doc/sourcebuild.texi (PowerPC-specific attributes): Add
2687         documentation of the powerpc_popcntb_ok attribute.
2688         * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
2689         code to issue warning messages if a requested CPU configuration is
2690         not supported by the binary (assembler and loader) toolchain.
2691         (spe_init_builtins): Add two assertions to prevent ICE if attempt is
2692         made to define a built-in function that has been disabled.
2693         (paired_init_builtins): Add assertion to prevent ICE if attempt is
2694         made to define a built-in function that has been disabled.
2695         (altivec_init_builtins): Add comment explaining why definition
2696         of the DST built-in functions is not preceded by an assertion
2697         check.  Add assertions to prevent ICE if attempts are made to
2698         define an altivec predicate or an abs* built-in function that has
2699         been disabled.
2700         (htm_init_builtins): Add comment explaining why definition of the
2701         htm built-in functions is not preceded by an assertion check.
2703 2017-01-04  Jeff Law  <law@redhat.com>
2705         PR tree-optimizatin/67955
2706         * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
2707         Allow any SSA_VAR_P as the base objects.  Use integer_zerop.  Verify
2708         the points-to solution does not include pt_null.  Use DECL_PT_UID
2709         unconditionally.
2711 2017-01-04  Uros Bizjak  <ubizjak@gmail.com>
2713         * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
2714         Use gen_int_mode instead of gen_lopwart for const_int operands.
2716 2017-01-04  Jakub Jelinek  <jakub@redhat.com>
2718         PR tree-optimization/71563
2719         * match.pd: Simplify X << Y into X if Y is known to be 0 or
2720         out of range value - has low bits known to be zero.
2722 2017-01-04  Alan Modra  <amodra@gmail.com>
2724         * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
2725         * configure: Regenerate.
2726         * config.in: Regenerate.
2728 2017-01-04  Jakub Jelinek  <jakub@redhat.com>
2730         PR bootstrap/77569
2731         * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
2732         a substring of the message, but strcmp with the whole message.  Ifdef
2733         ENABLE_NLS, translate the message first using dgettext.
2735 2017-01-03  Jeff Law  <law@redhat.com>
2737         PR tree-optimizatin/78856
2738         * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
2739         (mark_threaded_blocks): Remove code to truncate thread paths that
2740         cross multiple loop headers.  Instead invalidate the cached loop
2741         iteration information and handle case of a thread path walking
2742         into an irreducible region.
2744 2017-01-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2746         PR target/78900
2747         * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
2748         assertions.  Add support for doing the signbit if the IEEE 128-bit
2749         floating point value is in a GPR.
2750         * config/rs6000/rs6000.md (Fsignbit): Delete.
2751         (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
2752         Update the length attribute if the value is in a GPR.
2753         (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
2754         the sign or zero extension instruction, since the value is always 0/1.
2755         (signbit<mode>2_dm2): Delete using <Fsignbit>.
2757         PR target/78953
2758         * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
2759         extracting SImode to a GPR register so that we can generate a
2760         store, limit the vector to be in a traditional Altivec register
2761         for the vextuwrx instruction.
2763 2017-01-03  Ian Lance Taylor  <iant@google.com>
2765         * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
2767 2017-01-03  Martin Sebor  <msebor@redhat.com>
2769         PR tree-optimization/78696
2770         * gimple-ssa-sprintf.c (format_floating): Correct handling of
2771         precision.  Use MPFR for %f for greater fidelity.  Correct handling
2772         of %g.
2773         (pass_sprintf_length::compute_format_length): Set width and precision
2774         specified by asrerisk to void_node for vararg functions.
2775         (try_substitute_return_value): Adjust dump output.
2777 2017-01-03  David Edelsohn  <dje.gcc@gmail.com>
2779         * doc/invoke.texi (RS6000 options): LRA is enabled by default.
2781 2017-01-03  Eric Botcazou  <ebotcazou@adacore.com>
2783         * doc/invoke.texi (SPARC options): Document -mlra as the default.
2784         * config/sparc/sparc.c (sparc_option_override): Force LRA unless
2785         -mlra/-mno-lra was passed to the compiler.
2787 2017-01-03  James Cowgill  <James.Cowgill@imgtec.com>
2789         PR rtl-optimization/65618
2790         * emit-rtl.c (try_split): Move initialization of "before" and
2791         "after" to just before the call to emit_insn_after_setloc.
2793 2017-01-03  Gerald Pfeifer  <gerald@pfeifer.com>
2795         * doc/md.texi (Standard Names): Remove reference to Java frontend.
2797 2017-01-03  Pierre-Marie de Rodat  <derodat@adacore.com>
2799         * dwarf2out.c (gen_enumeration_type_die): When
2800         -gno-strict-dwarf, add a DW_AT_encoding attribute.
2802 2017-01-03  Jakub Jelinek  <jakub@redhat.com>
2804         PR tree-optimization/78965
2805         * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
2806         Change first argument from const call_info & to call_info &.  For %n
2807         set info.nowrite to false.
2809         PR middle-end/78901
2810         * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
2811         possibly throwing calls.
2813         * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
2814         and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
2815         and fns handling, rather than in a separate case SSA_NAME.
2817 2017-01-02  Jeff Law  <law@redhat.com>
2819         * config/darwin-driver.c (darwin_driver_init): Const-correctness
2820         fixes for first_period and second_period variables.
2822 2017-01-02  Uros Bizjak  <ubizjak@gmail.com>
2824         PR target/78967
2825         * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
2826         (*insvqi_1): New insn pattern.
2827         (*insvqi_1_mem_rex64): Ditto.
2828         (*insvqi_2): Ditto.
2829         (*insvqi_3): Rename from *insvqi.
2831         (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
2833 2017-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
2835         * doc/cfg.texi (Edges): Remove reference to Java.
2836         (Maintaining the CFG): Ditto.
2838 2017-01-01  Jan Hubicka  <hubicka@ucw.cz>
2840         PR middle-end/77674
2841         * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
2842         transparent aliases.
2844 2017-01-01  Jan Hubicka  <hubicka@ucw.cz>
2846         PR middle-end/77484
2847         * predict.def (PRED_CALL): Update hitrate.
2848         (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
2849         * predict.c (tree_estimate_probability_bb): Split CALL predictor
2850         into direct/indirect/polymorphic variants.
2852 2017-01-01  Jakub Jelinek  <jakub@redhat.com>
2854         Update copyright years.
2856         * gcc.c (process_command): Update copyright notice dates.
2857         * gcov-dump.c (print_version): Ditto.
2858         * gcov.c (print_version): Ditto.
2859         * gcov-tool.c (print_version): Ditto.
2860         * gengtype.c (create_file): Ditto.
2861         * doc/cpp.texi: Bump @copying's copyright year.
2862         * doc/cppinternals.texi: Ditto.
2863         * doc/gcc.texi: Ditto.
2864         * doc/gccint.texi: Ditto.
2865         * doc/gcov.texi: Ditto.
2866         * doc/install.texi: Ditto.
2867         * doc/invoke.texi: Ditto.
2869 Copyright (C) 2017 Free Software Foundation, Inc.
2871 Copying and distribution of this file, with or without modification,
2872 are permitted in any medium without royalty provided the copyright
2873 notice and this notice are preserved.